Research on behavior learning for intelligent robot based on evolutionary algorithms

To improve autonomy and adaptability of robot in dynamic real environment, a method which introduced evolutionary techniques into automatic design of adaptive robots is promising. The obstacle-avoidance behavior learning by genetic algorithms and genetic programming is implemented. Adding leaf node's mutation to make algorithms jump out of local-solution and increase convergence speed; the multi-points evolution and evolving multi-maps can successively enhance adaptability of solution. Finally, the result is compared by genetic algorithms and genetic programming. The simulation result shows the improved genetic programming is effective.