Science & Enterprise logo
Science for business people. Enterprise for scientists.

Engineers Develop Robotic Driver-Assist Safety System

Driver-assist test vehicle (Sterling Anderson, MIT)
Driver-assist test vehicle, with camera and range finder mounted on top (Sterling Anderson, MIT)

Researchers at Massachusetts Institute of Technology have devised a semiautonomous safety system using robotics to help cars and drivers avoid accidents. Sterling Anderson, a PhD candidate in mechanical engineering and Robotic Mobility Group researcher Karl Iagnemma presented their findings last month at the IEEE Intelligent Vehicles Symposium in Madrid, Spain (paid download required).

The system built by Anderson and Iagnemma uses a camera and laser rangefinder built into the car to identify hazards in the vehicle’s environment. An algorithm they developed then analyzes the data received and identifies safe zones in which the car can drive.

The system allows a driver to control the vehicle, only taking over control of the wheel when the driver is about to exit a safe zone. Anderson calls the system an intelligent co-pilot that monitors a driver’s performance, and makes adjustments behind the scenes to keep the vehicle in the safe zones or takes control of the vehicle to avoid collisions.

The MIT system differs from similar automotive innovations, such as self-parking cars or completely autonomous vehicles. The assumption behind these developments is the vehicle must follow a predefined route. “The problem is,” says Anderson, “humans don’t think that way.”

Rather, say the researchers, humans identify a continuing succession of safe regions in which to drive, such as an open lane on the road ahead. Using that approach, the team applied the algorithm with input from the camera and range finder to identify safe zones shaped as triangles, in which the car could proceed.

The edges of the triangles marked the outside boundaries in which the driver could navigate the car. If the vehicle leaves the safe-zone triangle, or if an obstacle enters the safe zone — such as a construction barrier — the system takes over, steering the car back into the safe zone.

One advantage over a fully autonomous self-driving vehicle noted by auto industry experts is the more modest technology required for a driver-assist system. Self-driving vehicles require many more expensive sensors and vast amounts of computation to work.

The team has run more than 1,200 trials of the system in Saline, Michigan with robotics and biometrics developer Quantum Signal LLC. The researchers say the system has encountered a few collisions, mainly from glitches in the vehicle’s camera that failed to identify an obstacle. For the most part, the system has successfully helped drivers avoid collisions.

Anderson points out the tests indicate drivers generally do not notice the system working in the background, which for most drivers is not a problem, but for new drivers could give them an inflated sense of their driving abilities. Likewise, expert drivers could find the system constraining. He and Iagnemma are now exploring ways to tailor the system to various levels of driving experience.

In the following video Anderson tells more about the driver-assist system.

Read more:

*     *     *


Posted

in

by