VW Cooperates with Mobileye and Valeo on future ADAS (L2+)
VW is planning to enhance safety features in future MQB platform vehicles in collaboration with Mobileye and Valeo. Coupled with its recent announcement with Rivian it appears that VW is moving further away from dependence on its internally developed CARIAD software. Sourcing hardware and software together and streamlining procurement across multiple brands allows VW to reduce complexity and improve efficiency according to Dirk Grosse-Loheide of VW’s Executive Committee. Future ADAS features will include hands-free driving (in certain conditions), traffic jam assist, emergency assist, parking assist and augmented reality displays. Valeo will provide the (single, central) compute unit hardware and sensors while Mobileye contributes its Surround ADAS™ platform, including the EyeQ6 processor.
Mobileye’s Surround ADAS system is a subset of its SuperVision system that offers full point-to-point navigation including urban pilot capabilities. Surround ADAS is intended to run on a single ECU structure and address lower cost points.
Honda sells 50,000 ADAS Equipped Cars in India
Honda introduced its ADAS system in India in May 2022 in the Honda City. The Honda SENSING system uses a forward looking camera for object and lane marking detection and performs Collision Mitigation Braking, Road Departure Mitigation, LKA and auto high-beam. The 50,000 unit number represents approximately a 30% adoption rate since 2022.
Uber Working with 14 Autonomous Companies – Globally
Uber rolled out its service with Waymo in Austin TX in March and after starting its own robotaxi development in 2015, is now working with 14 autonomous companies to integrate services into its app, including sidewalk delivery companies Avride, Cartken and Serve Robotics. The deal with Waymo involves Uber managing the fleet with services including cleaning, maintenance and charging and Uber set up a separate subsidiary to do that. Uber also handles the pricing and routing .
Hyundai partners with Avride to develop robotaxis
Avride was a spin off from Yandex (Yandex was the Dutch parent company of the Russian internet giant) and is working to integrate its technology into the Hyundai IONIQ 5 that will be rolled out in Dallas later this year in conjunction with Uber. Hyundai was also the primary backer of Motional, in a JV with Aptiv that is still working on L4 technology, but Aptiv decided to no longer allocate capital to that venture last year. The Avride system uses five high-resolution lidars along four radars, eleven cameras and an Nvidia compute platform.
Wayve trains its autonomous driving software in the US
Wayve has started training its self driving software in the US. Every region has unique driving scenarios (like right turns on red in the US), and Wayve trained its fleet with 8 weeks / 500 hours of driving on urban streets and highways. Wayve is backed by Microsoft and uses Nvidia compute platforms. Its software is based on a “foundation” model which can learn from a broad spectrum of unlabelled data. The fact that the model can be retrained with minimal data is key to keeping development costs down. They can also adapt the software to new vehicles with minimal training.
Renault Partners with WeRide
Renault is partnering with WeRide on autonomous electric minibuses. Renault is deploying driverless shuttles in Valence, Barcelona and Zurich airport following its trial last year at the French Open and projects demand for thousands of these buses over the next decade. Nvidia was an early investor in WeRide and WeRide believes they will solve driver shortages in aging societies. The partnership with Renault is one of their first entries into Europe.
ZF Testing L4 Driving in Germany
ZF received approval in Germany to start testing L4 vehicles on public roads. The latest shuttle holds up to 22 passengers and can operate on defined routes with mixed traffic. The ZF ProAI computer delivers 1500 TOPS and uses lidar, radar and camera sensors. The software was developed in partnership with Oxa.
Nissan Testing Autonomous Driving in Japan
Nissan has started its first public autonomous driving tests in Japan using a Nissan Serena Minivan equipped with 14 cameras , nine radars and six lidars mounted on the roof. Nissan plans to launch a service in 2027 as part of its Ambition 2030 plan. The technology is developed in house. Nissan has also been testing outside of Japan, in London and other areas of the UK. Nissan has been collaborating with Luminar on the lidar sensors.
Plus and Tier IV partner in Japan
Tier IV is addressing the truck driver shortage in Japan by partnering with Plus. Plus has worked with Traton – Scania/Man/International in Europe and the US to release advanced beta versions of its SuperDrive software last year and has been mapping hub to hub routes in Texas in preparation of commercial deployment. They have also worked with Hyundai and Iveco. It remains to be seen exactly how Plus and Tier IV’s open source platform are integrated together. Tier IV also achieved certification for L4 bus service (between the airport and station) in Komatsu, Japan.
Pony AI Scales Up Its Robotaxi Fleet
Pony has more than 300 robotaxis in China and plans to expand the fleet by more than 1000 this year, get to 10K by 2028 and become profitable by 2029. Pony also secured a permit for paid driverless operation in Shenzhen. Pony plans its own AV (with SAIC) with a target cost of $40,000 including the sensors and compute. Tesla has targeted a lower price point for its vehicle, but so far has not done any large scale public testing. Pony uses both Luminar and Robosense lidars.
Zoox resolves braking issue as it prepares for commercial robotaxi operations
Zoox had to recall 258 vehicles to resolve an issue that was causing the robotaxis to brake hard when a motorcycle approached quickly from behind or a cyclist entered a crosswalk. This was done as a software update. Separately, NHSTA has ruled that robotaxis without rearview mirrors and brake pedals (like Zoox) do not meet current safety requirements for on-road operation, which will either require them to add these features or petition for an exemption.
REE Automotive signs MOU to produce thousands of autonomous vehicles.
The deal, with an unnamed global tech company leverages REE’s P7 modular electric truck platform powered by its own zonal compute architecture capable of over the air updates. REE had been working with Navya on Level 4 autonomous software. Navya is based in France and was started in 2014, delivering its first and last mile passenger transport solution in 2015.
From the DMW to operate its robotaxi with a safety driver but would need a separate permit to operate driverless.
Tesla is behind Waymo and other operators in getting the required state licenses for robotaxi services and it will take some time to catch up.
Tesla Robotaxis service approval in California
Tesla was awarded the approval from the California CPUC allowing the company to own and control a fleet of vehicles and transport employees on pre-arranged trips which is a pre-requisite for starting robotaxi service.
Waymo to expand to Washington DC
Waymo announced that it will expand operations to Washington DC in 2026, using its own Waymo One App. Waymo is working with the city to change the rules requiring a safety driver. It’s estimated that Alphabet is spending around $1.5B a year on Waymo R&D and they raised $5.6B in late 2024 to expand service. It’s not clear at what scale this becomes profitable but I would estimate this happens at a fleet size in the low 10K’s.
Xaiomi Fatal Accident
A Xiaomi SUV crashed in China killing all three occupants. The car was apparently in NOA mode when it detected an obstacle, braked but then crashed into the side of the road at a speed of 97kmph. The accident gained widespread coverage on the internet in China with people starting to question Xiaomi’s “intelligent driving” and the company’s advertising claims.
It is going to be impossible for driverless cars to avoid all accidents or to prevent drivers using their systems not-as-intended, however high profile crashes like this will slow down adoption of these technologies and OEMs need to be more careful on how they promote use of the systems.
Stars Technology to use Seyond Lidar
Stars is developing self-driving container moving platforms and plans to use Seyond lidar as a core perception unit. More than 800 million containers are moved each year and the specialized equipment required to do this is another market for autonomous driving and advanced sensors.
Caterpillar to use Luminar Lidar
Caterpillar will integrate Luminar’s Iris lidar in its next generation autonomous solutions. The off-road trucks will use two Luminar Lidars. While the off-road market is relatively small these trucks can cost up to $5M and the cost of the hardware for autonomous operation is relatively small compared to the safety benefits it brings.
LiDAR versus Tesla Tests
Mark Rober put out a YouTube video showing some tests of a Luminar lidar equipped vehicle versus a Telsa Model Y. A simple AEB test at 40mph passes with the lidar, but fails with the Telsa (in manual drive mode) but passes in Autopilot. Next he tries the AEB on the dummy in fog. Again, it passes with the lidar car, but fails with the Tesla Autopilot. A third test uses simulated rain in front of the dummy, again the lidar car passes and the Tesla fails. Then he tries driving into blinding lights. Both cars pass that test. Finally the most controversial test, with a painted “wall” that looks like the road continues. Of course, the lidar picks up the solid object, but the Tesla fails – this is not a very realistic test though.
