New research answers one of the industry’s biggest questions: how many self-driving cars can one human safely keep an eye on?
For years, experts have debated just how many driverless cars a single person can effectively supervise from a remote control centre, with estimates all over the map. Research from Coventry University has finally identified a “Goldilocks zone,” showing that the sweet spot is between five and seven vehicles.
The findings are necessary for the safety and reliability of upcoming services like driverless buses, delivery vans, and robotaxis, where one operator will need to watch over a whole fleet of vehicles on set routes. This doesn’t apply to private self-driving cars just yet; for those, a driver still needs to be in the car and ready to take the wheel.
Published in ‘Computers in Human Behavior,’ the study makes it clear that getting the number right is a serious matter. Assigning too many or too few vehicles comes with life-threatening risks. The results offer a much-needed baseline for companies building the remote operation hubs that will act as the backbone for future transport, helping them boost efficiency, cut costs, and reduce the risk of accidents.
Researchers at Coventry University’s Research Centre for Future Transport and Cities set up their experiment to avoid the confusion of past studies. They created a simulator that looked just like a future control room and had 24 experienced drivers take on the role of supervisors.
Participants in the research study watched fleets of three, five, seven, and nine self-driving cars navigating a highly realistic “digital twin” of Coventry. Their instructions were simple: don’t interfere with the driving, just watch. If something looked wrong, their job was to alert a separate standby driver to step in.
While loading up an operator with nine cars might seem efficient, performance dropped off a cliff. At that level, supervisors missed or completely ignored over a third of the critical incidents that required a human’s attention. Their average response time held steady at around 13 seconds – about how long it takes someone to properly size up a situation before acting – but their overall awareness and decision-making suffered badly.
More surprising was the discovery that being underloaded is also a problem. When operators only had three vehicles to watch, many became bored and started to micromanage, intervening far too often when the system was working just fine. “With 3 I felt my attention wandering as there wasn’t so much to focus on,” one participant said. Another simply called it “dull”.
The sweet spot, according to the research, was supervising between five and seven self-driving cars. Here, operators found the perfect balance. They were alert and quick to react but didn’t feel overwhelmed. Clear information from the cars helped them make good decisions, though they noted that too many text messages became a distraction. The researchers believe that other ways of delivering information, like audio or voice alerts, could be a great solution.
Professor Stewart Birrell, Director of the Research Centre for Future Transport and Cities, said: “We’re proud to have provided insights that could support the safe rollout of connected and automated vehicles. This study was essential for understanding what remote supervision might look like in the real world. We found that five vehicles are enough to keep operations efficient without overwhelming the human operator.
“Having this knowledge and insight is critical if we want these systems to work in busy places like cities, where one person might need to keep an eye on several vehicles at once.”
This research is already being put into practice through the government-funded SCALE project, which is set to deploy remote control systems for self-driving cars in Solihull.
(Photo by Tuan Nguyen)
See also: Atlas robot learns new tricks using human-watching AI model
Want to learn more about IoT from industry leaders? Check out IoT Tech Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
IoT News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
