Introducing AV4EV: Our Initiative to Study Commercialization of Autonomous Driving

The automotive industry is on the brink of a massive transformation as autonomous vehicles (AVs) and electric vehicles (EVs) are poised to experience widespread growth and adoption over the next several years. While these new technologies hold tremendous potential in reducing carbon emissions, improving safety, enhancing mobility, and increasing efficiency, there are still significant challenges to overcome from the business side of AV and EV adoption. 

One of the most significant challenges facing the industry is how to develop and commercialize these new vehicles. The automotive industry is divided between traditional incumbents like Ford and GM, who have cash and infrastructure but tend to move slowly, and new challengers like Tesla, Rivian, and Lucid, who are more agile and inventive but also more financially volatile. To prepare for the changes of the next decade, the industry must learn to overcome this divide and embrace knowledge-sharing and collaboration. 

To address this challenge, the Mack Institute’s Program on Vehicle and Mobility Innovation joined forces with the Safe Autonomous Systems Lab at the School of Engineering, and the Autoware Foundation to launch a new initiative: Autonomous Vehicles for Electric Vehicles (AV4EV). The initiative consists of student projects aimed at fostering a global innovation ecosystem around the Autoware Foundation’s open-source technology for autonomous driving. Professors Rahul Mangharam from the School of Engineering and Mack’s John Paul MacDuffie lead the student projects.

“Autonomous driving has the potential to provide transportation solutions to areas and populations that are currently underserved,” said Christian John, Vice Chair of Autoware’s Strategic Planning Committee and President of Tier IV North America. “For instance, in Japan, where our company operates, there is a shortage of drivers to support the aging population that cannot drive themselves. Autonomous technology can help address these unique transportation needs for the elderly, such as going to the hospital or the store.”

“Secondly, the goal of Vision Zero, where there are zero traffic accidents and fatalities, could be achieved through autonomous technology in the long term. And finally, there is the significant benefit to the environment that can come from applying autonomous technology to transportation, such as optimizing routes for better fuel and energy consumption.” 

The Big Question: Open-Source Software (OSS) vs. Proprietary Models 

Over the Spring ‘23 semester, the AV4EV working group launched three projects focusing on a major debate in the automotive industry: the open-source software (OSS) versus the proprietary model of commercialization. 

Open-source software (OSS) is a licensing model under which the software’s original code is freely available and may be redistributed and modified. This model emphasizes collaboration and transparency, motivating multiple contributors to work together to develop and improve the technology. Proponents say the OSS model is associated with lower costs and increased innovation, but critics say it simply doesn’t work in practice—people “take” from the OSS without giving meaningful contributions in kind. 

The proprietary model is a licensing model under which a single company retains exclusive ownership and control over the intellectual property associated with the technology. This model emphasizes control and monetization, as companies can charge fees for access to their technology and maintain a competitive advantage over rivals. Making money is more straightforward with this traditional model but critics say it can lead to a lack of collaboration and interoperability, which can slow innovation and hinder industry-wide progress. 

“There is still a need for a lot of innovation in AV technology, including the computer platforms and sensors used in autonomous vehicles,” said John. “It’s unlikely that a single company can sustainably make the necessary investments and progress alone. Instead, an open architecture approach where multiple companies in the ecosystem innovate and contribute their part to an open-source technology is more promising. This is similar to what happened in the enterprise market, where open architectures, open-source software, and third-party integrations ultimately prevailed over complete, vertically integrated solutions.” 

About Our Spring ‘23 Projects 

What Is the Best Model for an OSS-Based Startup? 

In one project, students are working with the Turkish startup LeoDrive, which creates turnkey business solutions for autonomous driving. As a startup, LeoDrive has historically relied on the open-source model, which enables it to develop solutions more quickly, but which prevents it from generating recurring revenue via software licensing. Our students are using LeoDrive as a use case to assess the tradeoffs between the open-source and proprietary licensing models and propose profit-generating business models for an OSS-based startup. 

Vitality of an Open-Source Community 

Another project assesses the health and performance of an existing OSS community —that of the Autoware Foundation. Autoware currently has hundreds of contributors who believe in the power of open source. However, it struggles with questions common to many OSS models: how to reward those who contribute greatly, minimize “free-riding,” and prompt traditional businesses to switch from a proprietary to OSS model. Our students are compiling and analyzing data on consortium members’ contributions to Autoware’s open-source software and the benefits they derive from it in order to see if the community as a whole upholds the give-and-take principle of OSS development. On the basis of this analysis, we will explore interventions that might strengthen the community and increase the attractiveness of the OSS model.  

Software-Defined Vehicles 

Our final project involves Arm, a leading semiconductor company committed to a “software-defined future” in which vehicles will continuously improve via over-the-air (OTA) software upgrades. This project looks at SOAFEE, an industry-led initiative to create “cloud-native” software architecture for vehicles. Since cars are becoming increasingly dependent on powerful, cloud-based software and vast amounts of data, it is imperative to have efficient, safe, and secure connections to the cloud if widespread OSS adoption is feasible. As with the project above, our students are working to determine a business model for SOAFEE that maximizes the benefits of OSS and minimizes its risks. 

Looking Ahead 

As we move into the summer months, the AV4EV group is looking to the future. The next few years promise to bring major changes for the automotive industry which will open new opportunities for academic research. While fully autonomous cars on public roads are a high-hanging fruit, autonomous electric vehicles in “closed systems,” such as hospitals, airports, and university campuses, might be just a few years away.

“Our objective is to foster around Autoware’s open-source software a thriving global ecosystem that benefits from essential contributions made by Penn researchers,” says Dr. Valery Yakubovich, Executive Director of the Mack Institute. “Presently, we support interdisciplinary collaboration between Penn Engineering and the Wharton School. However, advancing autonomous and electric vehicles necessitates breakthroughs in other domains, including design, regulation, and insurance, which opens research prospects for Penn schools of Arts and Sciences, Design, and Law.”