Teleoperation and Orchestration Critical to Current and Future AV Fleets
September 1, 2020 | Company Blog
In 2017, General Motors predicted that it would deliver “thousands” of self-driving vehicles to create an all-electric, autonomous ridehailing service. GM was not alone. Companies from Alphabet (Google) to Zoox were angling for what seemed like a hugely profitable market. Taxi-like services that eliminate the expense of a driver, which is about 88 percent of the service’s cost, held great promise.
Needless to say, 2018 came and went without the mass arrival of autonomous fleets. And we’re still waiting. Self-driving technology just hasn’t evolved fast enough to allow vehicles to operate under “all conditions” — the threshold for the legal deployment of fully autonomous vehicles.
Meanwhile, regulators allow AVs on public roads so long as a “safety driver” is on board as a backup. Of course, the safety driver requirement defeats much of the purpose of driverless mobility services as the cost of the non-driving safety driver isn’t much different from the cost of an ordinary driver.
The requirement is not likely to change soon. “Despite substantial recent progress by the industry, fully automated driving systems that have no safety driver onboard will take at least a decade to deploy over large areas, even in regions with favorable weather and infrastructure,” the MIT Task Force on the Work of the Future reported in July.
While the safety driver requirement persists, a new wrinkle is that the drivers don’t have to be inside the vehicles. In many jurisdictions, “teleoperators” — humans working in central command centers — are allowed to serve as safety drivers with the training and technology to take the wheel via remote control.
Teleoperation offers a bridge not just to the deployment of individual autonomous vehicles, but to the delivery of mobility services using driverless cars. With individual teleoperators monitoring multiple vehicles, the economics of service delivery improves dramatically. Getting full-fledged services on the road can also allow service providers to better test and refine AV technology based on real-world performance. Teleoperation can help operators “achieve a safer, viable solution faster,” Amit Rosenzweig, CEO of teleoperation provider Ottopia Technologies told Autonomous Vehicle Technology.
Service delivery and fleet performance can be further enhanced and understood by marrying teleoperation and fleet orchestration technologies. Fleet orchestration technology like Bestmile’s enables service providers to optimize service performance by crunching demand and trip data to always deliver the most efficient ride available. Orchestration is also critical to the economics of mobility services. Making as many miles/kilometers driven carrying a paying passenger as possible requires processing trip requests, real-time and predicted demand, and traffic data to send the optimal vehicle for each requested ride (or delivery).
Bestmile and Ottopia have announced a partnership to integrate their teleoperation and fleet orchestration platforms to enable fleets to factor in human interventions in routes in real time. Say, for example, a vehicle on its way to pick up a passenger encounters an accident that blocks the road. A teleoperator takes over and drives around the accident. Instead of delaying the pickup, the booking automatically switches to a more convenient vehicle, and the assignments for the entire fleet are updated to compensate for the delay, notifying all affected passengers or delivery recipients.
Teleoperation and fleet orchestration will be needed not as temporary bridge to safe, efficient autonomous mobility services, but a permanent one. Even as self-driving technology improves, there will still be edge cases and circumstances with a human may need to take over. Extreme weather, technical malfunctions, unexpected obstacles the sensor may not understand, will still need to be accounted for.
Also, as services and fleets scale, monitoring vehicles and efficiently matching ride requests with cars becomes more and more complex. While we tend to think of autonomous services a single vehicle picking up a passenger, which might seem simple, we should instead think of hundreds or thousands of passengers booking trips simultaneously, perhaps with pooled rides for added efficiency. Matching travelers with vehicles, delivering the most efficient experience, and optimizing the utilization of the entire fleet requires vast computational intelligence.
Teleoperation and fleet orchestration play different roles in what Intel’s Mobileye has labeled the “Mobility as a Service Technology Stack.” The “stack” concept recognizes that while driverless technology is, of course, crucial to the delivery of autonomous vehicles, operating sustainable services with these vehicles will require multiple technologies working together — teleoperation and fleet orchestration being two of them. Both technologies play critical roles in enabling the safe, efficient, operation of autonomous fleets now and in the future.