The Foxconn-backed MIH “open source” EV platform project took significant steps towards being able to offer a complete, modern vehicle solution by forming an alliance with the AI experts at Autoware Foundation. The move will help accelerate the development of an integrated ADAS solution for the platform.
It’s a move that makes sense for Foxconn and its “Mobility in Harmony” (MIH)/Foxtron project to provide vehicle design and execution services to other carmakers and e-mobility startups. Up until now, it “just” offered an electric “skateboard” chassis. And, while that’s more than enough to put a vehicle on the road, electric car buyers are largely premium car buyers, and they expect a certain amount of autonomous capabilities from their vehicle. By bringing in the AI experts at Autoware, an open source platform in itself, the company can now offer boutique/would-be carmakers a proven and capable autonomous driving solution.
What is Autoware?
Founded in 2015 by Shinpei Kato at Nagoya University, Autoware Foundation’s open source AI platform is supported by one of the largest autonomous driving open source communities, with 2,300 stars on GitHub and over 500 accounts (claimed) on Slack. It’s an active community, too. As I type this, they show two repositories having been updated in the last six hours.
The Open AD Kit provides a reference framework for Software Defined Vehicle (SDV) development in commercial autonomous drive solutions. Similarly dense is Autoware’s description of the Open AD Kit initial release, which reads:
The initial release of the Open AD Kit includes cloud and edge solutions from ADLINK, Arm, AutoCore, AWS, SUSE, and Tier IV, as well as the implementation of Autoware supporting the containerized microservices architecture.
What that basically means is that lots of entities – from tech schools and universities to race teams and OEMs – are using some version of Autoware’s AI in their ADAS development projects. And, because it’s open source, progress and improvements made by one party can benefit all parties involved.
As for who these parties are, some of the bigger names include LG, Toyota (through TRI-AD, a research group focused on autonomous driving), Tata, TomTom, ARM, Apex.ai, and the US Dept. of Transportation. An array of other ride-hailing, robotaxi, and autonomous bus startups are also involved, and they all seem to be focused more on “mobility as a service” products vs. anything like “Level 5 autonomy”.
For what it’s worth, everyone involved in the Foxconn/Autoware alliance seems enthusiastic about its prospects. Jack Cheng, CEO of MIH Consortium said “To enable developers to innovate and create great user experience and applications, we must work towards a Software-Defined Vehicles future by realizing the Open EV Platform vision. MIH is excited to join force with the Autoware Foundation in the area of autonomous driving.”
Autoware founder Shinpei Kato echoed that view, stating that, “MIH will define the requirements for automotive grade AD solutions, and when automotive grade solutions based on Open AD Kit become available, they will be integrated into the AD/ADAS of the MIH Open EV Platform.”
Is Autoware any good?
As with many open source projects, it’s tough to pin down exactly what’s going on at Autoware, and how much progress is being made behind-the-scenes. That said, Autoware was behind a number of teams at The Indy Autonomous Challenge, which set autonomous speed records last October when they took part in the first high-speed autonomous racecar competition at the historic Indianapolis Motor Speedway. And, if that event was a reliable indicator: it’s pretty good.
The AI’s capabilities wowed audiences at a second Indy Autonomous Challenge event held during last week’s CES show in Las Vegas, as well, when teams took to the track at Las Vegas Motor Speedway and put on a high-speed thriller that included this:
That’s the TUM team’s self-driving IndyCar recovering from a 169 mph spin with – well, if not as much skill as an IndyCar driver, certainly more skill than you or I (statistically speaking).
They may not have won the race, but with that recovery? They won something, I promise!
Electrek’s Take
This is a smart move from Foxconn that allows it to really offer a competitive EV/autonomous vehicle solution for small-volume manufacturers. All the same, with Foxconn and Geely seemingly best of friends lately, after forming a JV in China to manufacture chips while collaborating on the vehicle platform-as-product business with the Zeekr-branded Waymo One, it’s a bit surprising to learn that Foxconn won’t have access to Geely’s excellent, Volvo-developed autonomous driving capabilities. That begs the question: why not?
Sure, it’s conceivable that the “open source” car platform might be best served, from a marketing or cost-control perspective, by an open source AI – especially one that’s probably familiar to the recent university graduates who will be helping to develop the next generation of self-driving cars. That’s possible, but I think the most likely answer is that Volvo isn’t willing to share.
Source | Images: TechTaiwan, MIH, Indy Autonomous Challenge.
Subscribe to Electrek on YouTube for exclusive videos and subscribe to the podcast.