Skip to Content

Autonomous urban car with predictive information

A commercial car, Citröen C3, was equipped to be a AGV. The only sensor used is just a camera and special paint and light. The region of visualization of the system is limited by a black metal box in which are installed the visual system. This metal box is situated in front of  the vehicle and its dimensions are just 30 x 50cm, so the region of visualization is hardly restricted yet the car can reach speeds of 50kph in straight ways and 20-30 kph in depends of the radius of the curves. To control the autonomous car a Fuzzy controller with just 49 rules was developed using the MOFS (Miguel Olivares' Fuzzy Software), obtained great results like reach high speeds and complete a tour of more than 4 km.

 

 

Here is the control loop of the vehicle, where the visual information give two different kind of data, the position of the line in the image and the code information when it exists. The code info goes to the offset block taking into account the speed of the car, the time elapsed since the code detection and the curvature ratio of the curve, giving offset value to turn in each curve. The Fuzzy controller use the visual information of the position of the line and the measure of the speed to generate an absolute position of the steering wheel. Also exists an integrator part out of this controller that gives more information of the line position, this integrator part has a little constant that reduce so much the power of this part, being active just with low speeds.