A new video from Wayve, a London-based autonomous vehicle startup, shows how its technology can help an autonomous van navigate the streets of London, England.
Wayve has developed a single general driving model that is capable of driving two vehicle platforms with different geometries. First, the system can be used on autonomous light commercial vans and second it can be used with autonomous passenger vehicles.
The company is working simultaneously on the technology for both platforms, which is tricky considering both types of vehicles have different mass, braking systems, geometry and sensor positioning, acceleration and width/length on the road.
For the autonomous van, Wayve chose a Maxus e9 light commercial vehicle that it adapted its autonomous vehicle system to. The AI model was the same that it ported to the passenger car versions and kept the same autonomy stack and vehicle abstraction layer.
In the video, Wayve’s autonomous van takes two trips through the public streets of London, encountering different types of traffic during both trips.
See the types of London traffic our autonomous van encountered during the first few days of on-road driving. During the first trip, the autonomous van must navigate through an area with pedestrians and cyclists on the road and through an entire block that has been blocked off for construction. Also in the first trip it has to navigate along tight streets with parked vehicles in the path.
The second trip for the autonomous van follows the same route. This time the van does not have the same cyclist traffic, but the roadwork is still in place as well as the parked cars.
Wayve said the two trips show how not every trip is the same despite being on the same route and autonomous vehicles must deal with different scenarios constantly even if the vehicle has mapped out the path ahead of time.