MIT Smart Cars Predict Collisions

By Kevin Lee

1,016 views Jun 15, 2011

Follow Kevin

intelligent transportation system

We've seen cars with lane departure warnings and collision detection, but MIT is pushing the envelope with tech that can predict human driving patterns. The computerized intelligent transportation system (ITS) uses an algorithm created from modeled human driving to predict where human drivers are going to go next.

ITS is also equipped with a system of onboard, roadside—which means States will have to build sensors into the roads—and traffic-light sensors to fortune-tell future conditions. While the computer can help a driver with stability control and traction assists, the car is not completely controlled by the computer. Instead it acts as an early warning system that uses beeps and lights to alert the driver of vehicles in their blindside or unapparent dangers.

Thankfully the system is not paranoid, so drivers won't be warned of a theoretical crash at every intersection. ITS uses an extensive model of driving patterns and situational awareness (whether the car is at an on ramp or turning into a one-way street) to determine what a car will do and acts only where there is a worst case scenario.

So far, MIT has only tested ITS in laboratory tests using model cars. The test consisted of one model car fitted with ITS and another controlled by a human controller, the two cars raced on circular overlapping tracks. Of the 100 trials, ITS only encountered the possibility of a collision three times and there was only one instance where contact actually occurred.

[MIT]