A camera-based cruise control system that automatically slows a car down to avoid potential accidents has been developed by Spanish researchers.
The prototype system 鈥� which uses a single dashboard camera to monitor traffic ahead 鈥� has been installed and tested by Miguel 脕ngel Sotelo and colleagues at the University of Alcal谩, Spain.
The onboard cruise-control system uses the camera鈥檚 video feed to identify road markings and determine road speed, then scans the road for other vehicles.
Advertisement
Once identified, other vehicles are monitored in case they come too close to the car carrying the system. The prototype, developed by Sotelo is connected to the car鈥檚 accelerator and brake and can slow it down, or even stop it completely, in response to cars ahead.
Sotelo says the most challenging issue is teaching the computer to distinguish vehicles from other objects and also to identify new models of car. This is done using a machine-learning technique, called support vector matching, to 鈥渢each鈥� the computer what visual characteristics to associate with an automobile.
Low visibility
Some car makers have shown an interest in the system, Sotelo says, but remain to be convinced of its viability. 鈥淲e have spoken to several manufacturers but haven鈥檛 broken through,鈥� he says. 鈥淲e鈥檙e still working on it.鈥�
Sotelo admits it will be crucial to improve the overall reliability of the system in order to persuade people it would work safely. Another key problem is that the system has trouble working in low visibility.
However, Sotelo says the visual system could be combined with another sensor, such as radar, in order to work in such conditions. Some car manufacturers 鈥� Jaguar, for example 鈥� have worked on radar-based cruise control in the past.
L茅onard Jaillet, a robotics expert from the Laboratory for Analysis and Architecture Systems in Toulouse, France, says the use of a single camera is clever. But he agrees that the system might well work better in combination with another form of sensor.
Sotelo presented his work on Thursday at the international conference on Intelligent Robots and Systems (IROS), in Sendai, Japan.
