A meetup at the Chinese electric vehicle manufacturer Nio at their San Jose, CA location gave a glimpse into the planned self-driving technology. The panel included several engineers from their self-driving team, who were available to questions from students of Udacity’s Nanodegree for Self-Driving Engineers.
First Nio showed a video from March 2017, where the NIO Ep9 races on the Circuit of the Americas with an average speed of 160 miles per hour and set a record for an autonomous(!) electric vehicle. The team behind the self-driving feature of the EP9 took not more then 3 months with 3 people to accomplish this impressive feat
After this impressive demo the engineers gave us a glimpse behind the scenes, led by Senior Director for Autonomous Driving, Jamie Carlson. The team today has 35 people and is expected to grow to 70 by the end of the year. In total there are 310 employees at the San Jose location. The space opened in September 2015 with 10 people and went through a strong growth. The biggest NIO location is in Shanghai with 2,000 employees. Also Munich has a design team with 80 employees, several of which worked formerly for BMW on the i8.In 2020 NIO wants to sell a level 4 capable autonomous electric vehicle. At the beginning level 4 will be available for highways only, but later software updates shall slowly enable autonomous level 4 city driving.The built-in sensor comprise the usual package of cameras, lidars, radar and ultra sonic sensors. The team is also looking at new sensor types that are coming from interesting startups, but Jamie could not yet share more details on that.
Nio does not believe that vehicle-2-infrastructure (V2I) or vehicle-2-vehicle-communication (VsV) is important. Today no other cars use that, therefore this has no priority for the team.
Furthermore Jamie does not believe in the sharing-model for NIO, as he as a father knows how it is having children seats and toys all over the car, and he can’t imagine him cleaning out a shared car all the time. NIO is designing a car for ownership, not for a sharing-model.
The video above also showed how ‘behavioral cloning’ was used to prepare the car. A human driver raced the track and the computer copied the behavior. This won’t be possible in production cars, as city driving has too many changing scenarios.
After the panel the evening continued with discussions and networking at the buffet.
NIO’s vision looks like this: “give people time back – to be who they want to be”
This article was also published in German.