Previewed by the ID.Buzz concept inherited from the past, Volkswagen’s reborn bus will arrive in 2022 with some interesting tech tricks up its sleeve. It will be fully electric, it will run on the MEB architecture already present under electric vehicles like the ID.3 and ID.4, and it will spawn an autonomous shuttle which should start carrying passengers in 2025. Argo AI helps Volkswagen to teach the Bus how to behave, and TechToSee got a glimpse of the project.
Volkswagen unveiled the first ID.Buzz-based prototype on the sidelines of the 2021 Munich auto show. Entirely draped in camouflage to hide its final design, the van is equipped with an armada of sensors, radars, cameras, microphones and lidars that paint a digital picture of the world around it. Argo AI – in which Volkswagen and Ford jointly own a stake – argues that its technology is very advanced: its lidar can detect and avoid potholes by scanning the road surface, and it can see objects that come together. are about 1,300 feet away, even though they are dark (like a black car). Powering this hardware requires enormous computing power, multiple backup systems, and a huge amount of data.
Argo AI operates two test tracks. One is in Pennsylvania, where the company is based, and the second is near Munich Airport in Germany. On these gated private properties, engineers push prototypes to the limit in a wide variety of conditions, including tight turns with reduced visibility and flooded roads. Fake animals and pedestrians are pushed into a car’s path to see how it reacts, while fog generators do their best to disrupt the sensors. The main focus of Argo is security. Its technology must work in cities, rural areas, at night, in stormy weather, etc. In turn, the data collected through testing improves the software. The more situations he encounters, the better he drives.
“We have a clear idea: it’s not about getting the biggest dataset in the world, it’s about getting the most complete dataset in the world to ‘power’ the system,” Bryan Salesky, Founder and CEO of Argo AI, said during a presentation.
A fascinating part of developing stand-alone technology for global applications is that the software has to be adjusted for specific regions. For example, Parisian motorists have different driving habits from those in Miami, and the car must adapt to its environment; it is not the other way around and there is no one-size-fits-all solution.
Road rage is not part of the training of an autonomous car.
“Our basic algorithm is the same, and then we adapt it to the natural driving behavior of motorists in each city. Washington, DC has a lot of roundabouts, and we’ve learned that already there and it gives us a head start when we get to European cities, ”noted Reinhard Stolle, vice president of the engineering for Argo AI.
That doesn’t mean that Volkswagen’s autonomous prototypes will pick up on every city’s bad habits, such as speeding, double parking, or incessant honking. Road rage is not part of the training of an autonomous car. They aren’t programmed to use their horns, they avoid sudden maneuvers and drive exactly at the speed limit – for better or for worse.
“We know that a vehicle that respects all the rules, especially speed, is not very popular in general traffic. [in Germany]. For now, in the first step, it is mandatory to do this. What happens afterwards, when we really know that the technology is mature, we’ll see. Exceptions may be possible, ”Stolle said cautiously.
“Our set of sensors is impressive, but from a business standpoint, it’s a lot cheaper than a driver. “
Until then, Argo AI will continue to test its technology on track and on the road to collect more data. It is currently testing cars in six US cities and plans to start rolling out prototypes on the streets of Munich in the coming months. After the testing phase is complete, Volkswagen and Argo AI will work together to make self-driving vans available in Hamburg, Germany from 2025. Other cities will likely follow later in the decade.
Autonomous technology is still expensive, but riding an autonomous shuttle can be cheaper than hailing an Uber.
“Our set of sensors looks impressive, but from a business standpoint it’s a lot cheaper than a driver,” said Robert Henrich, CEO of Moia, owned by Volkswagen (a division created in 2016 to provide mobility services such as carpooling). He added that a self-sustaining shuttle doesn’t need to take a break (unless it’s charging) and can run around the clock if needed. “The business case improves dramatically once we replace the driver – human drivers are pretty inefficient by comparison. We expect prices to drop significantly for the end customer, ”he added.
While all of this looks promising, autonomous shuttles still have many hurdles to overcome before they can merge into the mainstream. Technology still needs to be perfected, infrastructure improvements are needed, and a regulatory framework that allows self-driving cars to share the roads with human-powered vehicles needs to be put in place. When it comes to technology, it’s not just the hardware and software that make a car drive itself that needs work. Companies hoping to deploy autonomous shuttle fleets also need to figure out how to monitor what is happening during a journey and respond appropriately and in a timely manner. What if a runner has a heart attack?
“We will have indoor cameras, we will do indoor surveillance and we will rely on artificial intelligence. During the journey, we monitor what each passenger is doing and, in an emergency, the remote operator can connect to the vehicle via video connection and fix the problem. If necessary, the vehicle will stop, ”noted Henrich. One of the advantages of autonomous shuttles is that hitchhiking will require creating a profile with personal information, just like providing basic information about yourself before signing up for Uber. In theory, this helps ensure that runners remain courteous. In application, the question is much more complex, and it is one of the gray areas still being worked on.
Passenger transport is only part of the equation. Scale is extremely important in the field of self-driving cars, and the technology platform that Volkswagen and Argo AI are developing will also power vehicles that deliver goods. Looks like the next Volkswagen bus will pick up where its predecessor left off – with a hint of high tech.