The teledriving initiative will use the data gathered from remote operated trials to launch more autonomy features gradually.
We’re hurtling towards a future where cars operate without human intervention. One startup is hoping to bridge the gap between human drivers and full AI autonomy with a teledriving initiative.
The effort acknowledges that full level 5 autonomy is taking a lot longer than people expected thanks to challenges in training, risk of mishandling, and public distrust of machines driving without human fail safes. The Berlin-based startup Vay will deploy what are essentially life-sized remote-controlled cars to get the public used to the concept of driverless vehicles and to provide further research in autonomy.
The deployment will provide driverless taxis. Users can order a taxi from a smartphone app, and teledrivers will navigate the car to the user’s location. Drivers operate the vehicle from a console, much like a video game. Users will take over to drive themselves to their desired location, and teledrivers will take over from there to return the car.
How teledriving navigates complex safety rules and regulations
Although driver consoles resemble video games, the company will vet and extensively train drivers to avoid the most common driving hazards: driving under the influence, driving distracted, speeding, and fatigue.
Drivers will be fully engaged in the car’s surrounding environment, and monitors will provide a full view of the entire area around the car. The company will use the data gathered from these trials and beyond to launch more autonomy features gradually, remaining focused on the efficient rollout of automation.
This human-machine collaboration will launch in Berlin in a limited test, but Vay plans expansion to the United States and Europe. The company is currently working on regulation compliance and solving the problem of supply-demand. The solution is expected to cost a fraction of traditional ride-share programs and offer riders more privacy with fewer interactions with strangers.