Someday, when droids walk among us, they’ll probably be polite pedestrians.
A new robot has the mobility of R2-D2 and the manners of C-3PO.
This knee-high, self-driving bot abides by social protocols as it weaves through foot traffic: keep right, pass left, respect others’ personal space. The machine, presented at the IEEE/RSJ International Conference on Intelligent Robots and Systems in Vancouver on September 25, paves the way for robots that could one day navigate sidewalks to make deliveries or transport people through hospital hallways.
Programming a polite mobile bot is different from designing a self-driving car, says Michael Everett, a mechanical engineer at MIT. Unlike the rigid rules of the road, “the rules that humans follow when they’re walking around are pretty loose.” Building a bot that’s light on its wheels among individuals with highly unpredictable paths is “a totally different ball game,” he says.
Everett and colleagues decked out their rover with webcams and a depth sensor, then taught the bot how to behave by running it through thousands of simulated scenarios. After several hours of training, the robot could cruise at walking speed down a busy hallway without tripping up anyone nearby.
LET’S ROLL This careful bot assesses its surroundings and adjusts its course accordingly every 0.1 seconds, giving it a smooth gait. MIT Aerospace Controls Lab
Y.F. Chen et al. Socially aware motion planning with deep reinforcement learning. 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems, Vancouver, Canada, September 25, 2017.
Y.F. Chen et al. Socially aware motion planning with deep reinforcement learning. arXiv: 1703.08862. Published online March 26, 2017.
M. Rosen. Zippy new jumping bot catches air again and again. Science News. Vol. 191, January 21, 2017, p. 12.
M. Rosen. For robots, artificial intelligence gets physical. Science News. Vol. 190, November 12, 2016, p. 18.
M. Rose. Humanoid robot tops other bots in defense agency’s challenge. Science News Online, June 9, 2015.
M. Rosen. Designing robots to help in a disaster. Science News. Vol. 186, December 13, 2014, p. 16.