November 30, 2011
Supervised Autonomy featured in TORC's Blind Driver Challenge
Recently, TORC Technologies received a lot of press coverage with their Blind Driver Challenge. To recap, a TORC by-wire SUV was outfitted with additional interface components that a blind driver could use to interact with the computer controller. Then, that blind driver was able to operate the car at speed down a racetrack.
While a person was in the driver's seat here, the blind driver challenge is clearly a major step towards wider deployment of unmanned systems in the future. A common research topic in unmanned systems is human assisted autonomy, or supervised autonomy, where humans'keep track of and interact with an unmanned system at a higher level.
For instance, in the blind driver challenge, instead of operating the gas pedal and steering wheel, the driver relied on the computerized SUV to identify obstacles and unsafe conditions. The driver then received a consolidated situational awareness of the surroundings through specially designed vibratory interfaces. With that information, he could decide whether to speed up, slow down, or maneuver to avoid collision. These commands were then given back to the computer in terms of x, which it translated into more detailed steering and throttle actuation.
We see supervised autonomy as a clear transition path along the road to fully unmanned systems. One or a handful of operators can oversee a large number of systems, only paying attention to them when situations of complexity or uncertainty arise. In turn, each of the unmanned vehicles may have only a limited amount of intelligence, but sufficient to be independent for minutes at a time.
While a person was in the driver's seat here, the blind driver challenge is clearly a major step towards wider deployment of unmanned systems in the future. A common research topic in unmanned systems is human assisted autonomy, or supervised autonomy, where humans'keep track of and interact with an unmanned system at a higher level.
For instance, in the blind driver challenge, instead of operating the gas pedal and steering wheel, the driver relied on the computerized SUV to identify obstacles and unsafe conditions. The driver then received a consolidated situational awareness of the surroundings through specially designed vibratory interfaces. With that information, he could decide whether to speed up, slow down, or maneuver to avoid collision. These commands were then given back to the computer in terms of x, which it translated into more detailed steering and throttle actuation.
We see supervised autonomy as a clear transition path along the road to fully unmanned systems. One or a handful of operators can oversee a large number of systems, only paying attention to them when situations of complexity or uncertainty arise. In turn, each of the unmanned vehicles may have only a limited amount of intelligence, but sufficient to be independent for minutes at a time.