A few years ago if you were pitching remote driverless vehicle control, known as teleoperation, you would have been laughed out of the room by most investors claiming that a large multi-billion dollar company was about to ship 100% autonomous driving within a matter of weeks. That was 2017.
There have been a number of learnings over the last few years around driverless systems, but there is still another reason to get laughed out of the room — direct teleoperation with a steering wheel.
Teleoperation with Steering Wheels is a Bad Idea for Driverless Vehicles
One of the most dangerous pieces of hardware in the driverless vehicle space is the “gaming” steering wheel. Why? The steering wheel tempts engineering teams to try and remotely operate a vehicle over cellular connections claiming it to be the solution to unblock driverless vehicle use cases. In reality, while remote steering the vehicle makes for a cute parking lot demonstration for an investor, it is neither safe nor scalable as a long-term driverless solution. While the cost of bandwidth can be mitigated, many teams are attempting this on vehicles without critical safety features designed into the chassis.
Over the last few years, I’ve had the privilege of leading amazing engineering teams working on driverless solutions from light electric vehicles to Class 8 semi trucks. A key learning was to avoid the use of steering wheels for teleoperation outside of very specific use cases at walking speeds. While the steering wheel was useful for parking lot or yard operations, it quickly became dangerous at speeds above 10–15 mph.
The problem with using the steering wheel is latency (i.e. delay) over a cellular network connection. Humans can get good at steering a vehicle with a 50–100 millisecond latency, but it’s extremely tiring and stressful for any length of time.
Many early teleoperation companies realized the latency problem, so they set out to solve the wrong problem — improving cellular connections. The result was years of research into multiple modems, connection optimization, mapping coverage areas, etc. This was akin to realizing your head hurts when you bang it against a wall… So you work on making softer walls. Or, you could question why you were banging your head against the wall in the first place?
In the case of vehicle teleoperation, you need to ask yourself why a human should be responsible for remote steering a vehicle and having any level of responsibility for collision avoidance.
Buttons, Not Steering Wheels
A fundamental learning of trying to scale a driverless fleet with teleoperation is that you need to have a sufficient level of autonomy in the system to reduce the workload for the human operator. Another critical lesson from past projects pertains to safety. If you truly want to develop something safe, you have to have all of the safety features handled by computers on-board the vehicles (e.g. lane keeping, collision avoidance, emergency stops, etc.). Human operators need something equivalent to pressing buttons to authorize behaviors, not turning a steering wheel or stepping on a pedal.
At Faction, we’ve developed a third generation teleoperation system that we call TeleAssist®. TeleAssist was developed as a complete driverless system with cooperative autonomous features to reduce the workload for human operators. All safety-critical features are handled on board the vehicle — humans are never responsible for pressing a brake or halting a vehicle in an emergency scenario.
A side benefit of having simple interactions is that they can be recorded and used for evolving our autonomous driving capabilities. By applying machine learning to the TeleAssist interactions, we’re able to develop task specific automation to further reduce operator workload.
Next generation teleoperation workstations are extremely cost-effective compared to the remote controlled, wrap around monitor, pseudo-gaming approach. The workload is drastically reduced for human operators by having a simpler control design, which allows them to be able to comfortably take a job assisting vehicles for more than a couple of hours.
Don’t throw away that gaming steering wheel just yet, though… Keep it for the video game console in the living room. 😉
Follow our Medium publication to stay up to date with Faction's recent blog posts.
Ain McKendrick is the founder and CEO of Faction which develops driverless solutions based on light electric vehicles. The company believes the future of sustainable transportation is to develop driverless vehicles that are safe, cost-effective, and right-sized to serve a range of use cases for both business and passenger transportation needs. For more information visit www.faction.us
