As a final project for Prof. Joe Paradiso's course at the MIT Media Lab, Sensor Systems for Interactive Environments, I designed and constructed a low-cost collision detection system for lightweight vehicles based on sonar sensors with colleague Nicholas Pennycooke. We employed Parallax Ping sensors to construct a multi-input array of eight sonar sensors that used TTL pulses to communicate with a central Arduino-based microcontroller host board. The microcontroller then passes the digital proximity measurements to a host computer via USB, where the data are visualized in Processing sketch. The sensitivity of the visualization was linked to input from an analog tachometer, constructed with two permanent magnets affixed to a DC motor and a digital Hall effect sensor to measure changes in the magnetic flux density resulting from revolutions of the motor. In practical applications on a lightweight electric vehicle, this tachometer would provide speed information from the vehicle hub motors and increase the sensitivity of the collision detection system at higher speeds.
The system has a range of approximately three meters, a resolution of under 3cm and successfully demonstrated - at an experimental level - that sonar sensing is viable as a "first alert" collision detection system for autonomous vehicle operation.
The project report containing research goals, images, and conclusions is available here (6 MB PDF).