Open innovation in big-data driven autonomous driving : exploratory study : open innovation in a world of intellectual land grabbing

(1)

Amsterdam Business School

International Management Track

Open Innovation in Big-Data Driven

Autonomous Driving

Exploratory Study

Open innovation in a world of intellectual land grabbing

Author: Caper Wauters Supervisors: M. Paukku, 1st Supervisor UVA

Student number: 11671408 V. Scalera, 2nd Supervisor UVA

Keywords: Big Data, Autonomous Driving, Collaboration, Open Innovation, Second Enclosure Movement, Privatization, Closed Innovation, Intellectual Land Grabbing.

(2)

Preface

First of all, I would like to thank Dr. Markus Paukku for his insights and supervision during this Master thesis. Secondly, many thanks to all interviewees, for their willingness to participate and for their time. Lastly, in advance I would already like to express my gratitude to Dr. Vittoria Scalera for being the second reader of my thesis.

State of originality

This document is written by Wauters Casper who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

The introduction of Big Data and AI in the quest for the first autonomously driving car, has a disruptive effect on the car industry’s way of innovation. The increased complexity seems to lead to new types of innovation models for the automotive market.

Will the Automotive industry evolve their innovation process from being a closed innovator to becoming an open one? Original Equipment Manufacturers (OEM’s) have historically invested in internal R&D to boost their innovativeness, as well as buying up companies to add knowledge to the firm. The need of increased innovation (speed) and cross-industry knowledge for developing autonomous cars (Enkel & Gassmann, 2010), forces the automotive industry to look outside its own boundaries as a firm and industry, to escape a cross-industry technological innovation dilemma.

The emergence of open innovation models and collaboration, offer a solution for importing cross-industry knowledge into the innovation process, closing the knowledge gap between new high-tech market entrants and the OEM’s. Then again, are these incentives for open innovation being thwarted by the sensitivity of the knowledge in a high technology driven market? In a market where data is power, the habit of privatizing collected data can counteract the open innovation movement. Is this Second Enclosure Movement of “intellectual land grabbing” overpowering collaborative efforts in the autonomous market space?

This research looks if automotive business models are evolving towards more intermediate forms of open innovation, with an increased emphasis on collaboration and information sharing and moving away from the ivory towers, where the entire innovation process is developed internally.

(4)

1 Introduction

The first big disruption in the transportation of people dates back to 1908, when Ford introduced an automobile that many middle-class Americans could afford. Back then everyone still got around on horse and carriage, but the model-T Ford revolutionized the transportation industry (Wikipedia, 2017). Now autonomous (self-)driving transportation is at the forefront of the next major progression regarding the automotive industry.

Once again, the automotive industry is an area ripe for disruption. Not only are there sustainable demand-driven changes happening in the form of electrification of cars, but more

importantly, there is the emergence of autonomous driving. Apple CEO Tim Cook notes this

as a major disruption looming in the automotive industry (Bloomberg, 2017). According to Cook there are three revolutionary vectors of change happening in more or less the same time-frame: self-driving vehicles, electric vehicles and digital ride-hailing. Of the three, autonomous driving seems to be the key driver to the next innovative revolution and everyone is trying to get in on this technology.

1.1 The Automotive Playing Field

Due to the fast innovation and increasing role of technology and data in the 3 converging revolutions described by Cook, very different business models can be expected in the future automotive sector. For the first time in 100 years the traditional automotive industry is being

shaken up, where legacy car manufacturers or Original Equipment Manufacturers1 (OEM’s)

face strong new opponents who are disrupting the entire automotive ecosystem (Bratzel, 2017). These new high-tech players like Google, Uber, Nvidia and Apple are already starting to

(7)

penetrate the automotive sector, experimenting with new business models in this new data driven playing field. Car manufacturers like Tesla are trying to internalize the entire value chain, while other firms are looking at collaborating and are trying to find their niche in this new space (Marshall, 2018).

A host of automotive brands and other tech heavyweights have been investing heavily

in autonomous R&D2, all searching for an effective business model in which innovation and

data are key elements. Tesla Inc. and Uber have used the three vectors of change to fuel the development of autonomous driving, it being via electrical vehicles or digital ride-hailing. Tesla uses its connected network of cars to gain insightful knowledge by collecting customers driving data, while ride-hailing companies like Uber and newcomer Lyft use their taxi customer base as a starting point for developing self-driving cars. Within the playing field of the autonomous automotive industry there are more and more competitors popping up by the day. Lyft has been on a streak of corporate collaborations, recently pairing up with autonomous

driving tech companies Waymo3_{and NuTonomy, to accelerate the deployment of autonomous}

vehicles. Furthermore, it acquired significant investment from Jaguar Land Rover in addition to the initial $500 million General Motors gave in December 2015 (Hawkins, 2017).

The traditional players in the automotive industry have been increasingly interested in mobility and technology companies, as they seek to insulate themselves from the possible decline in personal car ownership. By 2025, auto manufacturers are predicting that at least half of today’s drivers are unlikely to want to own a car (Rawlinson, 2017).

All these partnerships and fierce competition have not been without challenges. A recent example being that Uber was sued by Waymo for allegedly stealing autonomous driving secrets (Davies, 2017). Even though the case is still ongoing, this has not been without

(8)

casualties, including the head of Uber’s self-driving unit, accused of stealing 14,000 document from Waymo (Carman, 2017). This case shows the importance of knowledge and data in the new automotive playing field and although firms are collaborating, the degree of knowledge sharing can vary greatly.

Figure 1 -Twitter post Hemal Shah - source: The Verge 2017

1.2 Autonomous Driving and its Cross-Industry Complexity

Producing an autonomous car is such a new and complex innovation that to do so, a company will need a wide range of capabilities. These can be summarised in 3 key aspects; it needs to be knowledgeable in the manufacturing and distribution of cars (like traditional automakers), it must be able to process Big Data in order to develop the autonomous Artificial Intelligence (AI) (software start-ups like Waymo) and finally there has to be knowledge and access to the customer base. The kind of data Lyft has through its ride-hailing network. As market players don’t have all the needed resources for building a full-fledged autonomous car, an innovators dilemma arises with cross-industry knowledge gaps. The biggest challenge being how to bring high-tech knowledge and car manufacturing knowledge together. Currently there are car manufacturers that know how to build a car and scale production, but know little about big data collection, processing and AI training with “deep learning”. While on the other hand, you have these new High-Tech market entrants that poses a lot of knowledge about data AI and sensors, but have never built cars or managed a fleet.

(9)

How market players try to bridge their knowledge gaps can differ. Some believe in the more traditional closed innovation approach, where they build on their own strengths and internalize the lacking knowledge to their existing innovation by buying tech or collecting knowledge/data and privatizing this. The latter, closing off knowledge by privatization, is also known as knowledge enclosure. Most closed innovators take this very protective stance in fear of knowledge-spillovers to their competitors. At the same time, others see open innovation models as a way of sharing information and solving knowledge gaps, plus increasing their innovation abilities and speed. Fusing innovation together unlocking new insights and pushing the autonomous market as a whole, with less paranoia about knowledge-spillovers.

The strategy for how firms deliver the autonomous driving experience vary widely. There are firms such as Tesla, trying to offer the full package, from car to AI, for the autonomous solution. On the other hand, there are modular innovators such as TomTom and Nvidia, who focus on filling in a piece of the technology stack and by doing so excel in a niche in the value chain of developing an autonomous car. Decisions in this regard greatly influence the degree and necessity of cooperation between firms.

1.3 Research and Contribution

Technological developments have completely disrupted the automotive industry. The increased complexity of developing a car that can drive autonomously has pulled new High-Tech players into the market. Both the old and new market players are now caught up in the race to get ahead, testing their (new) innovation models in the process. In this high-tech playing field, a tension has arisen between the closed and open innovation models, however very little literature can be found that studies both open and closed innovation and none can be found that apply the theories on the autonomous car test case. Further literature research will create a better understanding of the 2 business models and how they interact.

(10)

This results in the following research question:

In the currently emerging autonomous driving industry, which innovation models are being adopted due to high-tech influences?

By conducting semi-structured interviews with the emerging industry leaders in autonomous driving, as well as attaining extra insight from experts in the closely related fields of Big Data, Open innovation and AI, an exploratory study is set up.

1.4 Thesis outline

The emergence of autonomous driving solutions is bounded by the use of Big Data for R&D of AI. To demystify the inner workings of autonomous driving some core concepts used in the development of autonomous driving will be clarified. The second chapter will start by describing Big Data as defined in literature. After boiling down the demarcations of core concepts like Big Data and AI for autonomous driving to showcase the complexities that drive the automotive market with new High-Tech inputs, the remaining literature review looks into the different innovation processes, starting by describing closed innovation and its evolving second enclosure movement, followed by a deep dive into open innovation. Next in chapter 3, a theoretical framework is made, where the relevance of this research will be further explained by formulating propositions about the innovation process of autonomous driving and by constructing a framework for categorizing a firms’ innovation approach. This will be summed up and visualized in a conceptual model. The conceptual model will be the starting point that leads to the research design in chapter 4.

(11)

2 Literature review

In this chapter, key concepts and processes influencing today’s technology and data driven automotive market are analyzed by means of literature, further clarifying complex topics. By elaborating on Big Data and the use of AI in autonomous driving, the sectors’ increasing complexity is sketched, highlighting the need of cross-industry knowledge to build an autonomous car. The resulting knowledge gaps due to these new complexities fuel the emergence of both fast paced collaborative innovation and Knowledge (En)closure. After analyzing the open and closed innovation models through literature an underlying tension between the two is signaled, making for an interesting business case.

2.1 Key Concepts and Processes in Todays Technological Market

2.1.1 Innovation shift

The dominant Resource-Based View in International Strategy states that a firm should strive to attain a set of unique core capabilities that can be leveraged to a competitive advantage (Barney, Wright, & Ketchen, 2001). According to Teece (1998), one of the key capabilities a firm must poses is a certain degree of innovative capability and the role of technology in innovation is ever increasing. For the development of autonomous driving, especially the mining of Big Data is generating new insights and opening up new innovation possibilities by introducing applications of AI (Chen, Chiang, & Storey, 2012).

The role of technology in innovation is resulting in new approaches to R&D. Traditionally firms kept their competencies and development internal, often with complete secrecy to the outside world, so that it’s R&D investments could be converted to competitive advantages and allow the firm a first mover position into a new market. Even though internal R&D is still a critical source to the innovation process, more firms are moving away from the

(12)

Trimi, 2012). This strategy change is fuelled by the fast pace of technological innovation. The landscape of innovation is changing at such a high pace, that the rate of development in innovation is a key factor, possibly overpowering the risk of knowledge-spillovers. It is no longer a question if, but when, a firm is able to develop innovations for new industries, with a lot more firms stepping into the technological development field. Look for example at the growth of firms in Silicon Valley alone (Abraham, 2017).

Figure 3 - Patent Registrations Silicon Valley // Data source: U.S Patent and Trademark Office

As seen in figure 3 Computers, Data Processing & Information Storage, all directly related to Big Data, are some of the newest but by far the fastest growing fields in Silicon Valley. These sectors facilitate new types of innovation due to data mining and machine learning, further speeding-up the rate of innovation change.

The new driving technological forces of Big Data and AI through machine learning, not only change the way innovation can be applied but also cause higher levels of complexity for the development of an autonomous car, generating a cross-industry knowledge gap between the High-Tech sector and car manufacturers. The complexities will be further explored by

(13)

2.1.2 Big Data and AI Driving Innovation Complexity

First of all, let’s take a closer look at this fuzzy term ‘Big Data’. It is a trending topic and nearly impossible to avoid in any new business opportunity. There seem to be many different definitions for Big Data and not without reason. Business is shaping up to be more and more data-driven and it is understandable that along the way the terms’ definition has been refined. Gartner’s 3V’s model is one of the first to define Big Data some 15 years ago. The definition endures until today, in which Big Data is characterized according to its Volume, Variety and Velocity (De Mauro et al., 2015). Big Data is large in volume, has a wide variety of data and a high speed of data streaming in. Beside the 3V’s, the definition of Big Data has been expanded, to include the need for a specific technology to capture and create value out of that data (De Mauro et al., 2015).

This is a good starting point from where to further specify what Big Data is and how its application can deliver solutions in the autonomous driving sector. Big Data functions as the key resource for developing machine/deep learning, which is crucial input for the autonomous vehicles Artificial Intelligence (AI). Applying algorithms to the Big Data, identifying patterns in the data and upon these insights predicting new patterns in upcoming data, creates an algorithmic system that learns by itself without further human interference (Hall et al., 2016). The availability of Big Data on human drivers’ behaviour is the most essential resource for the development of the car’s Artificial Intelligence (AI) and will determine the rate of innovation (Nothdurft et al., 2011; Thorpe, et al., 1991). Both studies from Nothdurft (2011) and Thorpe (1991) indicate that the biggest challenge in autonomous driving is the split decision making and anti-collision reaction of the AI by visual input. In order to create a car with autonomous driving decisions, specific Big Data is needed for the AIs ‘deep learning’, which comprises of discovering patterns so the cars’ AI can successfully

(14)

right decisions in its surroundings. The safety and success of an autonomous vehicle depends on what Big Data can be attained and what deep learning insights can be added to the algorithms of the AI in support of its decision making.

Here a crucial cross-industry knowledge gap arises. New High-Tech market entrants like Nvidea, Waymo, etc. are more knowledgeable about developing AI and processing big amounts of data, but lack the data mining capabilities due to a lack of fleet and inability to build cars as a whole. For legacy car manufacturers or even ride-hailing services it’s the other way around.

Thus, good user-data sets are crucial for the self-learning AI algorithms used in autonomous driving. It is impossible to pre-program all scenarios a car will encounter when driving on the road, so the AI must learn by repetition and replication (just like a human). Luckily, once it has seen a scenario it will never forget the lesson learned and make the same mistake twice, something that can’t be said for all people driving around.

2.1.3 Shifting innovation models

The previous paragraphs have described how in today’s technological marketplace, a paradigm change is happening in innovation due to complex cross-industry knowledge gaps. From literature two important innovation movements can be discerned: open versus closed innovation. On the one hand, there are industry leaders who resort to a resource based strategy for the accumulation of valuable technology assets and take an aggressive stance in protecting

their intellectual property. On the other hand, successful high-tech global players are seen

disconnecting the value chain to work together with part specific industry experts (Teece & Pisano, 1994), opening the doorway to open innovation models.

Within high technological markets like autonomous driving there are different forces pushing and pulling towards open and closed innovation models, creating a tension and

(15)

uncertainty of best practices within the automotive industry. Freeman & Engel (2007) argue that innovation requires two important underlying conditions. Firstly, the accessibility and mobility of resources and secondly, incentives must be aligned so that the collaborating parties (within a company or between companies) complement each other where needed. Current autonomous driving collaborations poses a lot of information asymmetry due to specialization and diversifications of collaborating partners. Within this organizational chaos the distribution of knowledge seems to be the main source of the problem in the form of knowledge-spillovers. The exploration of firms into new markets, coupled with the liberalization is stripping firm-level competitive advantages back to its fundamental core of innovation and making intangible know-how (or intellectual property, IP) the leading asset for competitive advantage (D. J. Teece, 1998). Pulling back open innovation initiatives towards closed internal innovation with higher IP protection.

To properly outline the market activities in terms of this innovation tension in the autonomous driving industry, a deep dive into the concepts of closed innovation and open innovation will follow.

2.2 Closed Innovation

2.2.1 First enclosure movement

Building on the tension between closed and open innovation models in the autonomous driving industry, are the phenomenon called the first and second enclosure movements that add contradictory logic. Closed-off innovation derives from a thought of protecting intellectual property (IP) by privatising it and enclosing it from the public. This concept is called the first

enclosure movement and it is a principle that dates back to the very beginning of economics

(16)

commons). The first steps towards enclosure where taken by people when they started to fence of pieces of these common grasslands for themselves. There was a positive side to this process, because unlike with commonly owned land, people were willing to invest in their own piece of land. This could be through fertilisation, irrigation, etc., increasing the production and efficiency of that piece of land. A fenced of area motivated people to invest, because there was a clear reward in the form of increased production from the specified piece of land. According to Boyle (2003), this first move of privatization made a strong case, the enclosure of property of the commons was a good thing and privatization leads to specialization and increased efficiency. To safeguard this movement, a development of property rights was needed to fuel progress and keep incentives for investment. Over the centuries people have applied this concept to almost all physical and even non-physical elements of life. Thus, privatization of intellectual property (IP) was born.

Safeguarding property rights for intellectual property is obviously not as simple as building a fence around it. The rewards of investment and creation of IP can be easily lost. In order to maintain the market incentives for innovation, protective steps must be taken for the creator/innovator. When unable to exclude others from copying their innovation, an innovator cannot get returns for their innovation investment. Thus, a limited monopoly must be created in order to reward the innovator and pay back his investment. This is called the intellectual property right. For intellectual information with low barriers and easy reproduction capabilities IP rights play a big role.

First Enclosure Privatization counteracting Open Innovation

Thoughts derived from this first enclosure movement still has lingering influences on the

development of open innovation. The first enclosure movement started as a positive movement,

(17)

efficiency of the privatized product and created trust for future investment. Especially in more software based platforms, privatization is the dominating strategy with a winner-take-all ideology, argued via logics of network externalities, natural monopolies and first-mover advantages (Eisenmann et al., 2010; Parker & Van Alstyne, 2005). The transition to more open, knowledge sharing models is still very limited and the progression of open innovation is yet

unclear. Ferraro and O’mahony (2004) discovered that even today, seemingly open innovation

projects still create (contradictory) organizational boundaries to safeguard the income from these projects, in some way still looking how to fence off certain pieces of the innovation process. From his research the question arises, do collaborative projects have a chance of remaining open at all, or do they funnel and narrow over time, becoming increasingly closed to a select few and eventually privatizing? While producing knowledge goods like software, a company can ensure security and stability of the knowledge flows by determining what to open and what to close off. But doing so handicaps the possibility for fully open innovation.

Further research indicates collective vs individual tensions arising in growing collaborative (open) networks (Wareham et al., 2014). As the ecosystem of a collaborative network expands in scale and scope, a contradictory tension is triggered in the form of a tragedy

of commons4. The growth and sheer size of the network can create contradictory logic and have

adverse effect on the behaviour of the ecosystems actors. Actors may decide to choose for their own wellbeing above the overall health of the ecosystem by say privatization, creating actual tragedy of commons. . As such effects multiply, the ecosystem risks becoming increasingly unsustainable and the same network that has enabled its growth facilitate its demise (Wareham et al., 2014). In other words, when actors of a collaborative innovation project start privatizing

(18)

new IP fuelled by the open collaboration, tragedy of commons may apply, for example when privatizing image processing algorithms built on collected data from the collaboration. These actions of self-interest risk triggering a closed attitude and counteract the possibility to openly share information in the collaboration.

These tragedies of commons due to enclosure are becoming increasingly visible in software/Big Data based solutions. Where firms go a step further by rushing to enclose ever larger stretches of potential information.

2.2.2 Second enclosure movement

In the software development sector, there is talk of the second enclosure movement (Boyle, 2003) and seeing as autonomous driving, for a large part, consists of software development and Big Data utilization, this is also applicable to this emerging industry. The second enclosure movement surpasses protective privatization and overreaches towards the likes of “intellectual land grabbing”! In a world where information/data is power, the sole purpose of this enclosure movement is fencing off future data sets and using potential open/common information. Firms are trying to utilize collaborative/open information, plus fence off the first adopters data in the autonomous market, by privatizing their products based on these information sources.

It can be argued that privatizing firms have a right to protect their collected data because they created the circumstances to harvest it and even though it is normal from a closed off innovators standpoint to protect IP regarding complex algorithms and computations, justification for the raw user data is less obvious. Firms even try to fence off future raw driver’s data without actually fully using the data right now. More focused on mining data and privatizing it then actual development for the future. Pulling knowledge and learning out of the (future) shared information pool. This second movement endangers the sustainability of open innovation projects as market incentives are destroyed and tragedy of commons apply.

(19)

In the new autonomous driving industry space, there is a risk that leading (high-tech) firms will try to internalize as much future information as possible by locking-in customers to their products and fencing off their user data. By doing so these firms could harvest potential future common knowledge or “intellectual land grabbing”, creating an advantage for the private firm itself but slowing down or even blocking some of the general generation of autonomous knowledge. As Boyle (2003) refers to this intellectual land grabbing like the environment, it is an information pool of commons that is yet to be discovered. “Like the environment, the public domain must be invented before it can be saved”. The question remains if this second movement of privatizing future data will potentially kill off all open innovation efforts in the autonomous driving scene.

2.3 Open Innovation

Open innovation is a powerful framework consisting of the employment, capture and generation of intellectual property (IP) at firm level (West & Gallagher, 2006). The model stresses the importance of using a broad range of knowledge sources for a firms’ innovation process, stretching from rivals, academics to even cross-industry firms. In the last decade, the concept of open innovation has been extensively researched and still it ranks high on the agenda for the management in technology and innovation. More research is demanded to gain an understanding of this emerging innovation management paradigm (Boscherini, Cavaliere, Chiaroni, Chiesa, & Frattini, 2009).

The open innovation paradigm is often contrasted to the traditional closed “proprietary” model by which internal R&D activities lead to innovative products (Chandle et al., 2009). Within closed innovation firms, IP that could not be commercialized can be licensed to others or “sits on the shelf” until internal development can put it to other use, reducing risk of ‘knowledge-spillovers’ to other firms (Chesbrough & Rosenbloom, 2002; Smith & Alexander,

(20)

The emergence of a more open model is presenting itself, where companies recognize not all innovation can come from inside the organization and the speed of innovation is increased by means of collaboration. First signs of an innovation transition from closed to open innovation models are emerging by a hybrid option, where even between competitors there is cooperation for certain parts of the innovation process also known as coopetition. By cooperating in parts of the value chain, certain innovation hurdles can be overcome to grow the market as a whole and still compete in other area’s of the value chain (Bengtsson & Kock, 1999).

The concept of open innovation is primarily applicable to the high-tech sector (Chesbrough & Crowther, 2006), with the automotive industry as a good example. According to a study by lli, Albers and Miller (2010), the open innovation movement was already present in the traditional legacy automotive industry. Table 1 shows the degree to which this was the case and outlines the two opposing innovation ideas. The table shows that open innovation already showed promise, but the tendency towards closed innovation prevailed. Nonetheless, the shift towards more open models appears at the horizon, with a more knowledge/technology intensive automotive market developing.

(21)

In the knowledge/technology driven autonomous driving industry, the shift from closed to more open innovation models seems to be happening right now. External sources of knowledge play a bigger role, complicating the evolution of early stage technology projects, involving technological and market uncertainties (Chesbrough, 2004). In these circumstances firm’s need to rethink their business models to better facilitate collaborative/open innovation, they are required to change the way they manage innovation. A firm needs to “play poker” in risk taking as well as play “chess” for strategic risk reduction. Strategic reconfiguration is needed to determine what knowledge to share and what to close off.

2.3.1 Advantages of Open Innovation

In todays increasingly high-tech environments, firms are required to reach outside the firms’ boundaries in order to access technological knowledge and enhance their innovation performance. Furthermore, it is often argued that a company actually starts to boost its’ innovation speed the moment information is openly shared. Cohen and Levinthal (1989) suggested a combination of internal and external R&D and in doing so serving a dual purpose. On the one hand this will boost internal development by complementing it with input of external knowledge and also this allows a firm to evaluate/observe innovation outside its own boundaries. In this context, firms with higher investment in their own (internal) R&D benefit more from external R&D and spillovers in future partnerships.

When using internal and external innovation, firms must be aware that this is not a one-way street. Once available, firms often assume that the sources of external innovations will continue to flow, but what happens when everyone in the partnership tries to be a ‘free rider’ by only absorbing external innovations? Firms start to show first signs of the Second Enclosure Movement. This behaviour is primarily linked to fear of knowledge-spillovers to a direct

(22)

competitor. Even though knowledge-spillovers can be problematic, studies have shown that it is still economically rational to keep sharing information and that open innovation generally pays off (Sofka An Grimpe 2010). Firms in the same industry complement each other in creating the market but compete in dividing it (Nalebuff et al., 1996). So, if a firm stands to benefit from an innovation that increases the market size, it will accept spillovers if the return from its share of market growth is attractive enough. Moreover, firms need to and seem to be willing to contribute back to the existing projects, to assure that the external innovations continue to meet their perspective needs, to maintain absorptive capacity, and to avoid discouraging current and future innovators for collaboration (West & Gallagher, 2006). A study by Henkel (2006) shows that the development of the Linux operating system is a good example of a successful open innovation platform. Firms utilizing Linux, contributed a lot of development back to the public embedded code, despite it not being directly in their best interest but in the best interest of the Personal Computer Marketplace. Similar for the autonomous driving industry, developing the market first by means of open innovation benefits all market players. If big steps towards autonomous driving can be made due to sharing information and the knowledge gaps can be filled, they will accept the risk of knowledge-spillovers.

Besides the boost it gives to a firms internal innovation and general market growth, another advantage of committing to open innovation is that it makes firms a more attractive partner with resource connections and expertise (Powell et al., 2005). The open innovation investment acts as a “ticket of admission” to the open innovation game. Only partners with considerable expertise, reputation, plus willingness to play together are invited. Resulting collaborative networks create more innovation by benefiting from technological discussion and knowledge exchange, which provides new opportunities and future business diversity (Lee et al., 2010).

(23)

What makes open innovation as a business strategy attractive, is the possibility of exploiting imported knowledge and building on idle intellectual capital, enabling even large corporations to become more entrepreneurial and increasing the speed of innovation (Dodgson et al., 2006). Open innovation models provide considerable efficiency gains, as it supports accumulation of knowledge (David, 2003) and reduces duplication efforts in the innovation process. In the case of autonomous driving the open innovation models overcomes industry knowledge gaps by sharing information between high-tech entrants and car manufacturers. Car manufacturers can share their user data, while the high-tech players can feedback the processed user data for say better imaging process algorithms.

2.3.2 Challenges of Open Innovation

Although there are a lot of potential gains when utilizing the open innovation model, it also bares considerable challenges. The costs of managing/coordinating the network of added external expertise are uncertain. Particularly as the number of interdependencies increases with more sophisticated and often competing demands on multiple relationships. Who bears the costs and which partner reaps the benefits from an open model still remain somewhat unclear. Open innovation is a powerful approach for generating extra innovation capabilities, but also can create intellectual Property (IP) issues at firm level. Thus, another source of uncertainty is concerning how to manage and protect a firm’s intellectual property (IP). Some firms address this issue by revealing selectively. For example, in programming language, this would encompass revealing most of the main source code without disclosing crucial add-ons on the main framework.

Beside the practical issues such as cost and methods for IP protection, the main challenge for most firms is that they are stuck in a proprietary attitude and their unfamiliarity with openness. Openness is new territory for most firms, that need to rethink their business

(24)

models in order to innovate. Based on innovation research by West and Gallagher (2006), 3 fundamental adaptations can be identified for applying the open innovation concept to a firm: exploit internal innovation, incorporate external innovation into the internal development and ensure information exchange both ways of the partnership. The latter creates a tension in most sharing models. Companies instinctively only want to receive the necessary information without disclosing much of its own knowledge (fear of knowledge-spillovers). This train of thought must first be broken before the actual full potential/benefits of collaborative/open innovation can be realized. Hughes and Wareham (2010) conducted a study focusing on building open innovation capabilities via external information sharing, creating uncertain knowledge boundaries in the innovation network. They discuss the importance a firm’s

absorptive capacity5 has in relation to the shared knowledge. Not only how to import gained

knowledge into the existing firms capabilities, but also the growth of absorptive capacity for the entire innovation network. The conclusion of their research was that firms must adjust to bi-directional knowledge sharing to some degree in order to embrace and utilize the open innovation model and enjoy the benefits of experimental learning and sharing of best practices. Exactly how much outbound knowledge must be shared is very industry and firm depended.

In the case of autonomous driving the question remains if legacy car manufacturers can switch their traditional proprietary mind toward open innovation without fear for knowledge-spillovers of IP, focussing more on the joined gains and increased innovation speed due to open innovation. Of course, there are hybrid transition methods by partial cooperation in certain parts of autonomous development, while competing in the rest (Khanna et al., 1998), but

5_{Absorptive Capacity: a firm's ability to recognize the value of new information, assimilate}

(25)

working so closely with your direct competitors only intensifies fear of knowledge-spillovers, potentially feeding the fear and limiting open innovation in other areas.

Of all challenges, how to distribute firm knowledge seems to be the biggest hurdle to overcome. The protection of IP keeps many firms at bay from open innovation.

(26)

3 Theoretical framework and Research Contribution

Available research and literature on the application of open innovation models to the highly complex, technological and data driven autonomous driving industry is very limited, if not

completely unavailable.Research is limited to the traditional automotive industry (Chiaroni,

Chiesa and Frattini, 2012) and more general research on open R&D and open innovation

((Enkel & Gassmann, 2010; Chesbrough & Crowther, 2006; D. J. Teece, 1986). Based on these sources and news articles the situation can be sketched of a fast changing industry, with large new players and varying business approaches. The technology push has created new innovation pressure, shifting automotive innovation structures to new forms and it is still unclear which business approaches will prevail. This makes the automotive industry an interesting case study for research of innovation models in modern technological markets, by combining the available generic literature and interviews of important market players. This chapter describes what contribution this research makes to a better understanding of the industries innovation processes. To start with a theoretical framework is constructed, which is used as an important tool for this research. Following this, the research question and several working propositions are presented.

3.1 Theoretical Frameworks

The concept of open innovation shows promising application within the automotive innovation shift, but can happen in many different forms and to varying degrees. Based on the concepts of closed and open innovation, a framework has been constructed to identify the type of innovation model each interviewed firm utilizes. By looking further into innovation literature and searching for existing categorizing frameworks, a base innovation framework was found. An adaptation has been made of the existing framework by introducing an extra hybrid option of open innovation, relating to more of a coopetition type of collaboration. Collaborating

(27)

openly in certain areas of autonomous development while competing in the rest with a more protective closed innovation model. In the following paragraphs the framework is presented.

Research from Haeussler (2006) indicates, the type of competitive relationship with your partners, as well as the knowledge sources and access channels, determines the level of knowledge control between collaborating partners. Furthermore, when the level of external knowledge inflow is considerable, firms also regulate the outflowing knowledge more loosely. If a firm acquires external knowledge from competitors by entering joint projects, more actions are taken to control outgoing knowledge. Hence, these firms try to establish rock solid fences around them. Contrarily, in an environment with high potential spillovers both ways, firms control the knowledge flows from a “take and give” angle instead of a “knowledge catcher” approach. Concluding, the extent of a firms’ collaborative innovation depends on the

management of the inflowing and outflowing knowledge. Based on similar theory, Ellen,

Gassmann and Chesbrough (2009) approach open innovation by way of knowledge flows. By categorizing the directions of information flow, they determine 3 broad archetypes of “open innovation”: inbound, outbound and coupled (bi-directional knowledge sharing), as illustrated in Figure 2.

(28)

Figure 2: Innovation Models by way of Knowledge Flows (Ellen, Gassmann and Chesbrough (Enkel et al., 2009)

Although the concept of knowledge flow identification for innovation model determination is a clear method, for this case study of open innovation in the autonomous driving scene, there is need of a concreter framework to be able to link firms’ structures to their innovation process.

West and Gallagher (2006) made a framework that describes knowledge flow structures related to the level of openness and collaboration. For the purposes of this report, the three innovation structures proposed by them have been expanded to include a fourth. Making an adaptation to the initial knowledge framework set-up by West and Gallagher (2006), results in the next 4 R&D knowledge sharing structures to identify the firms’ level of openness, see Figure 3. The additional structure is outlined in blue.

Scanning for new Techonolgies & Capabilities Development Pilots & Prototypes

R&D Configuration R&D Development

Internal R&D Technolgy Push External R&D (Cross industry Technology) Inbound

Process CoupledProcess Outbound

(29)

Figure 3: knowledge flows for increasingly open innovation models. An adjusted framework, initially by West and Gallagher (2006)

At the top, the closed innovation model refers to the traditional internalization of innovation. By developing R&D within the company without external innovation sources. Secondly, is the

modular innovation framework. This hybrid model was added because it was the missing link

that demonstrates the first step for a firms’ adoption towards open innovation. It’s a hybrid between closed internal innovation and open development within the innovation partnership or network. By adding the hybrid form based of the coopetition concept, the transitional phase is also covered. A firm can cooperate for a module of development to overcome knowledge gaps but still compete in the other modules of the developing value chain (Bengtsson & Kock, 1999). The firm can yield returns by combining internal and external technologies to offer a product otherwise not available. Not all member of the partnership share full information. Still showing signs of primarily external knowledge input (inbound knowledge) and trying to “freeride” on other partners. A lot of the R&D is divided into modules or silos, developing their R&D in parts, being a cog in the entire system. Information and knowledge sharing is kept to a minimum and only the absolutely necessary is shared, in order to protect their IP as much as

Firm R&D Closed Proprietary Innovation Hybrid Modular Silo Pooled Innovation Spinout Innovation Firm R&D Module 1 Firm Firm Firm R&D R&D Module 1 Module 2 Firm R&D Module 2 Firm R&D Module 3 Public Community Level of openness

(30)

possible. The range of openness can vary within this model from an absolute minimum to extra innovation sharing for increased innovation speed and insights. Thirdly, pooled innovation shares a lot more information. Knowledge is shared between partners of the innovation network. Here joined “pilot” projects of innovation form a central joint project open to all partners. Generated feedback, information and knowledge is freely shared within the boundaries of the partnership, to further optimize each partners’ contributing module. The speed of innovation is greatly increase compared to the more conservative hybrid silo framework. Lastly, spinout innovation projects are the true forms of open innovation. Where the innovation process and progress is actively shared to the public. By dumping the R&D into the public domain, it enables the community to freely access and track the growth of the R&D. This knowledge framework is the closest thing to actual fully open innovation. The Mozilla web browser is a good example of a public/community grown project in responds to the lacking capabilities of Internet Explorer.

Figure 4: Degree & Type of Collaboration (based on West and Gallagher (2006)) Closed InnovationClosed Innovation Modular (Silo) Innovation Competition C en tra l R & D D ecen tra l R & D Collaboration Pooled Innovation Spinout Innovation External Orientation Collaborative Orientation

(31)

Figure 4 plots the 4 innovation models by competitiveness and centralization of R&D, quickly giving a clear overview. The primary factor for choosing the openness of the partnership is based on knowledge-spillovers. Depending the risk and potential of spillovers for IP, firms choose their collaborative orientation and external orientation.

3.2 Research Question and working Propositions

Using the adapted innovation framework in Figure 3, an initial investigation can be conducted on how innovation is developing in the autonomous driving industry. By matching industry innovation processes to the constructed framework, development trends can be identified. Is there a transition from closed to open innovation models in the autonomous driving industry due to cross-industry knowledge gaps? Or is the risk of knowledge-spillovers to big, so firms show signs of enclosure movement and are counteracting possible open innovation? Once the dust settles which innovation model will prevail? As a result, the following research question is formulated to oversee the new research angle for the test case of autonomous driving.

3.2.1 Research question

In the currently emerging autonomous driving industry, which innovation models are being adopted due to high-tech influences?

To operationalize the research question, a conceptual model is built as a guideline for setting up the exploratory study, see Figure 5: Conceptual model research.

(32)

Figure 5: Conceptual model research

On the basis of this conceptual model, working proposition are made to help support the research question and facilitate direction during interviews, making sure to collect the correct feedback from the correspondents. The working propositions will work together with the innovation framework to discover firms innovation models and indicate how these innovation methods perform and interact in the market.

Working Proposition 1

Collaboration and sharing of information increases your partnerships’ innovation speed.

Openness and sharing of information between collaborating parties increases a partnership’s innovation speed as it supports accumulation of knowledge and reduces duplication efforts in the innovation process. But does this also apply to the autonomous market where the risk of

Firm Innovation Process Open innovation? Closed Innovation? level of knowledge openess level of knowledge openess Comparering to Innovation Framework Innovation Performance interviewed

firm innovation matched

(33)

knowledge-spillovers can be grave and we have already seen partnerships end in lawsuits over IP.

Based on quotes from the interviewed firms, their stance on the use of openness in innovation for facilitating increased innovation speed can be observed. By identifying if collaborative openness increases the innovation speed for the firm and if the rate of innovation has become prioritized above the risk of knowledge-spillovers.

Open innovation models prevail in the big-data driven autonomous driving industry.

Within the autonomous driving space there are different modes of openness towards sharing information about innovation development. This proposition says that the general consensus is that more open/collaborative business models generate better innovation result and perform better then parties that try to internalize all development.

With the innovation framework, the level of openness of each interviewed firm can be identified. Which type of innovation models is encountered the most, plus how do they perform in the autonomous driving market. Based on a recent report by Navigant Consulting (2018) a current standing of the autonomous development of firms is made and this can be compared to their innovation models used. Within the interviewed market players are there more open or closed innovators and who of them is taking the lead in autonomous development? Furthermore, interviews can indicate if a firm switched to a different innovation modes and for what reason.

(34)

The second enclosure movement is a real threat to open innovation.

Although the Second enclosure movement refers to the privatization of existing open data sets, this concept is still valid in the autonomous driving industry. In this case, the risk lies mainly in the information pool of commons that is yet to be discovered. “Like the environment, the public domain must be invented before it can be saved” (Boyle 2003). The premier bottleneck for autonomous driving is training the AI and for that great amounts of Big Data are needed. Closed innovators are actually performing “intellectual land grabbing” by committing customers to their product and with this lock-in effect removing potential future data sets out of the commons information pool. How much impact this movement will have compared to the rate of development of open innovators is to be seen.

Will the market become separated by companies chasing the silver bullet and achieving to be the ‘Apple standard’ in the market. Offering the entire product or will there be more of a modular separation where niches of the value chain get filled with a more Android like technology platforms and different companies running their cars on the platforms?

Based on the interviews, are firm moving away from collaboration efforts and refocussing on the internal closed-off innovation, like for example in the past Tesla and Mobileye split up after safety issues, see Appendix A. This was a while ago, but are there similar examples where firms have knowledge-spillover issues and reframe form openly sharing development? Furthermore, do the interviews show clues of firms actively focused on pure driver data collection with intent of shielding this off from the possible future openly accessible data pools, inhibiting open autonomous development in the process?

(35)

4 Research design

The level of analysis in this research is at firm-level. Big Data, Machine Learning and the Artificial Intelligence (AI) needed for an autonomous (self-driving) car is a crossover field that is newly developing. With the many cutting-edge initiatives, there is not enough viable data available for a quantitative research approach. Therefore, an exploratory qualitative approach has been used by interviewing the industry leaders and experts. Whenever possible, the interviews have been conducted in English for easier codification.

4.1 Research Structure

The development of the autonomous car industry was chosen as a specific case study for this research, so that the practical implications of disruptive technology and Big Data on collaborative knowledge structures can be clearly determined. This research makes an analysis of each firms’ different approach to autonomous innovation, within the specified developing industry, making it an embedded single case study (Yin, 2009). Both Yin (2009) and Zucker (2009) argue that in order to capitalize on the benefits of such a case design, there must be replication logic in either a literal or theoretical form (Zucker, 2009). In this case, a theoretical comparison applies. Theoretical replication logic looks at similar or contrasting approaches for the same innovation problems and the motivation behind these decisions.

(36)

The interviews conducted for this research are semi-structured, so there is a starting point in questions from where the direction of the conversation can be determined. The initial structure of the questions forms the basis of a comparison between interviews and enable linkage between interview insights. According to Saunders (2011) inductive methodology like an interview, see Figure 8, is a good approach to explore new theory/strategy and closely examine processes. To this end, the interviews were matched with the different innovation frameworks. By conducting the interviews, initial market developments for autonomous innovation can be drafted. During the interviews, patterns of innovation are observed and matched with the adjusted working propositions to eventually determine theoretical understanding of the markets innovation.

4.2 Interview Procedures

The interviews are semi-structured, forming the basis of the research instrument and allowing for the exploration of deeper motivations of interviewees, leading to higher quality answers and thereby achieving greater insights, as proposed by Horton et al., (2004). The interview structure, containing its questions, can be found in Appendix D. Autonomous driving is a reasonably unknown field and developing at a rapid pace. For this reason, (semi-)open questioning was deemed necessary. The interviews started from a semi-structured question followed by an open discussion on the topic. The structure provides a memory que for the interviewer and helps guide the conversation with follow-up questions if the discussion strays

Figure 6: Induction = Bottom-up approach

(37)

off topic. By recording the interviews, the interviewer could focus his full attention on the conversation. These recordings were later transcribed, to enable comparison between interviews through data analysis, a method prescribed by Pope et al., (2000).

Interviews were in English or alternatively in Dutch, depending on the interviewees preference and ability to communicate in English (judgement made by the interviewer. Whenever possible, interviews were conducted face-to-face to better understand the interviewees meaning by also reading their body language during the interview. Firms Amber Mobility and TomTom were interviewed in person. Due to interviewees time constraints or their location abroad, other interviews were conducted over Skype or Google Hangouts. The interviews varied in length, the shortest being 20 minutes with TomTom and up to 45 minutes with Amber Mobility. A second interview was conducted with Koen Lekkerkerker (RCS), because he indicated to have some new insights. To smoothen the conversation a switch was made from English in the first interview to Dutch in the second.

4.2.1 Pilot interview

To start with, a pilot interview was conducted at a small autonomous driving start-up from the TU-Eindhoven called Amber Mobility. Doing so enabled the interviewer to optimize the research design and practice his interview techniques. Making sure the structured questions point the conversations into the right direction and enabling the interviewer to cope with the flexibility and deepening on subjects from semi-structured follow-up questions and conversation (Kvale, 1996). The interviews were all recorded and transcribed, plus notes were taken during the interview itself. Probing during the conversation and recapping with a reflection or example helps structure the answers and minimised the risk of going off topic, as proposed by Leech (2002).

(38)

4.2.2 Interview pool

Multiple interviews were conducted from which data was collected and analysed. Most interviewees where approached though direct emailing or through the author’s own network. The interviewees are listed in table 1 below.

Table 2: List of interviewees, their organizations and functions

Name Organization Function

Frank Poort Amber Mobility CTO (autonomous/electric

ride-hailing service)

Vincent Laurense Stanford University PhD Researcher Autonomous

Vehicle Control

Vincent Demunynck TomTom Senior Product Marketing

Ryan Samaan Waymo (Alphabet/Google) Mapping Operations

Carlo van de Weijer TU-Eindhoven & ex-TomTom Reseach Track Automotive

Koen Lekkerkerker Robot Care Systems &

TU-Delft

Lead Engineer of WEpods self-driving shuttles

Tom Westendorp NVIDIA Sr. Business Development

Manager Autonomous Driving Koen Lekkerkerker (2)

follow-up Robot Care Systems & TU-Delft Lead Engineer of WEpods self-driving shuttles

Wiebe Janssen Ex-Telsa / Lightyear Powertrain mechanical

engineer at Lightyear / Ex-Tesla engineer model 3

Of the 9-interviewed market players, not all were free to openly share their thoughts and information. Especially from car manufacturers, there was a lot of hesitance to participate and restrictions during the interviews. An exception was Amber Mobility, who utilize more of an open, start-up culture with less fear of disclosing sensitive information. On the other hand, an example of a very closed attitude is that of a contact at Tesla (Tom van Rijndorp) who is responsible for AI development in Tesla’s autonomous program. He agreed to do an interview, but after checking with Tesla headquarters he was not allowed to continue. After further pursuit, it was possible to talk to an ex-Tesla employee Wiebe Janssen who used to be in charge of development of the model 3 Tesla car. However, he was very restricted in what he was able to share. With Uber there were similar circumstances, after speaking to 4 people at the firm

(39)

who were willing to participate, the Europe regional director closed down the conversation. Even though initial contact with leading car manufacturers in the autonomous driving industry was not successful, other big players were contacted and interviewed. The interview pool can be categorized into 3 groups: Research Institutes / High-Tech firms / Car manufacturers or ride-hailing services.

The universities of Stanford (Vincent Laurense) and TU-Eindhoven (Carlo van de Weijer) were approached in the category of Research Institutes.

- Carlo van de Weijer is in charge of the research track Automotive at the TU-Eindhoven and supports Amber Mobility plus other autonomous initiative. He is regarded as one of the leading experts in autonomous development in the Netherlands.

- Vincent Laurense does his PhD research in the field of Autonomous Vehicle Control. The lab where he does all of his testing is called the Center for Automotive Research at Stanford; CARS is the acronym, which unites many car manufacturers (OEMs), but also suppliers, car insurance companies, tire manufacturers to share ideas on mobility and solve automotive problems. Vincent is directly involved in a project where it works closely and openly together with Audi and Volkswagen.

Although getting through to car manufacturers actively participating in the development of autonomous driving was a big challenge, two parties were interviewed.

- First off, Amber Mobility shared their insights and innovation development for the future. At this time, Amber Mobility is a ride-hailing service with electrical cars that have semi-autonomous features. In the near future, they plan to no longer build on their BMW I3 test cars, but want to build their own car from the ground up, called the Amber One. Their motivation for building their own car is to better integrate their own

(40)

autonomous bridging systems, as conventional cars do not really enable fully autonomous support.

- Second, Koen Lekkerkerker of Robotic Care Systems (RCS) involved in autonomous shuttle bus WePod was interviewed. RCS is the leading firm in the collaboration for building the WePod shuttle bus. Initially, the collaborations started-off with many participants, but after problems with closed innovators the collaboration slimmed down to only actively open collaborators, like for example Nvidia.

- And Thirdly Wiebe Janssen an ex-Tesla employee was interviewed, but sadly the interview was not fruitful without any new information disclosure or insights. The aim of the interview was to circumvent the Tesla HQ shutdown of my interview with Tom van Rijndorp at Tesla, by interviewing an ex-Tesla employee and gain some insights of how they innovate, the company culture and his personal experiences. But during the interview he was reluctant to disclose even the smallest detail with regard to Tesla because of a signed NDA. This gives a good indication of how closed-off Tesla innovates and instils this into its (ex-)employees.

In the category of High-Tech firms entering the autonomous driving industry, Waymo, Nvidia and TomTom were contacted.

- The content driven mapping company TomTom was interviewed about how they plan to position themselves in the developing autonomous market and what degree of open innovation is possible for a company so heavily dependent on data collection.

- Nvidia is the leading computer chip manufacturer for autonomous driving systems. Beside their hardware, they offer an open software development platform for autonomous driving. Nearly all autonomous vehicle developers use Nvidia hardware and software, even fully closed companies like Tesla have switched to integrating their

(41)

system. As the computing power of the Nvidia chip and software is 10x bigger than anything else on the market right now.

- Lastly, Ryan Samaan in charge of mapping at Waymo was interviewed. Formerly Waymo was known as Googles self-driving car project that started in 2009. Now Waymo is the subsidiary of Alphabet focussing on autonomous car development. Waymo stated “we have no plans to become a vehicle manufacturer or supplier to the

auto industry. Instead, Waymo intends to partner with other companies to provide vehicle platforms while retaining control of the automated driving stack and providing mobility services to consumers”. In 2017, Waymo announced a partnership with Intel

to develop autonomous driving technology together and develop better processing. In conclusion, a good spread of market players was interviewed. Giving a good representation of different innovators in the autonomous driving industry. Talking to car manufacturers, High-Tech entrants and highly collaborating Research Institutes. The fact that research institutes like Stanford also were able to give big insights into their collaborations with big car manufacturers like Audi and Volkswagen compensate partially for the failed interview attempts with Telsa and Uber. Combining these insights with the car manufacturers from Amber Mobility and WePod covers the car manufacturing side of the market to some degree. Enabling an assessment of the field of how they innovate and how they perform in the race for technological development.

4.3 Strengths and limitations of research design

Autonomous driving is a reasonably unknown field and a small market, making for small sample sizes, further supporting a qualitative approach but making the rigour of the study strongly depended on how close to the actual developing source the interviewee is. This is a primary market research where not all leading experts were willing to cooperate or could not

(42)

be reached. Yet looking at the array of interviewees and their place in the autonomous driving space, shows important players with leading innovations structures that seem to overlap and show market conform developing practices of how to solve the autonomous driving problem.

For a more thorough assessment of the field, more developers must be interviewed. There is still room for better representation of big car manufacturer, but for an initial insight into market innovation the interview pool suffices.

(43)

5 Data and Analysis and Results

This chapter analyses and presents the results from the conducted interviews. Based on the literature derived working propositions stated in chapter 3, the interviews will be structured by codification. By coding the key inter-related themes, market behavior towards the propositions is interpreted and explains opposing views towards open innovation models. Further learning effects from market performance can be added to the innovation stance of each market player and under what circumstances.

First, the discovered inter-related themes and enclosing codes are identified and structured, by executing a word query analysis.

Second, the propositions are analyzed in comparison to the interviewed firms and their stance to the matter. This is done by presenting the support or resistance of the proposition in a table combined with factual quotes.

Third, a cross-table overview summarizes the important findings per interviewee with regard to the propositions to create a clear overview.

5.1 Data Analysis

The initial step of the data analysis is coding the imported transcriptions. Coding qualitative data consists of summarizing parts of the interview in words or short phrases, assigning attributes to the text (Saldaña, 2015). By doing so, comparisons can be made between interviews by looking at overlapping codes and pattern recognition. To process the transcribed interviews the coding is analyzed, recoded, annotated and eventually the basis is created for a summarization of the interview, as can be seen in Figure 7.

Open innovation in big-data driven autonomous driving : exploratory study : open innovation in a world of intellectual land grabbing

Amsterdam Business School