A Genetic Approach to Finding a Controller to Back Up a Tractor-Trailer Truck

Problems of control can be recast as requiring the discovery of a computer program (i.e., control strategy) that takes the state variables of a problem as its inputs and produces the values of the control variables as its output This paper describes the recently developed genetic programming paradigm which genetically breeds a population of computer programs to solve problems. Genetic programming begins with a population of hundreds or thousands of random computer programs and improves them from generation to generation using the Darwinian operation of fitness proportionate reproduction and the genetic operation of sexual recombination. The sexual recombination operation combines parts of two parental computer programs to produce new offspring programs, each parent being selected proportional to fitness, This paper shows, step by step, how to apply genetic programming to the four dimensional control problem of backing up a tractor-trailer truck to a loading dock. Genetic programming breeds an approximately correct computer program (i.e., control strategy) that successfully performs the required task.