Automatically synthesizing a planning and control subsystem for the DARPA urban challenge

To incorporate robots into society, they must be able to perform complex tasks while interacting with the world around them in a safe and dependable manner. The recent DARPA 2007 Urban Challenge made a step towards that goal by testing how well robotic vehicles can interact in an urban environment while dealing with static and dynamic obstacles and other cars. This paper uses the Urban challenge to demonstrates a general approach for automatically synthesizing correct hybrid controllers from high level descriptions. Here we create a planning and control subsystem for the vehicle that, if the information gathered by the sensor is correct, satisfies the requirements of the challenge for different dynamic environments. This approach automatically produces a system that is guaranteed to behave according to the traffic laws while interacting with other vehicles. Furthermore, it allows systems to be changed rapidly and easily thus reducing design time and eliminating human error.