摘要
进化算法是一种基于种群和进化的启发式全局搜索方法,通过创新与改进使其适用于解决机器人的规划问题。根据家庭服务机器人仿真比赛的特点设计了一种全新的编码结构——事件结构,这种结构具有归一化、信息保留完整、便于操作等特点。使用代价计算法作为进化算法的适应度函数,并改进算法的进化方式,采用异位自交叉和限定变异法,从而使进化算法能够用于家庭服务机器人的任务规划之中。通过实验和比赛,验证了算法的有效性和优越性。
Evolutionary algorithm （EA） is a kind of heuristic global optimization technique based on population and evolution. It has been made applicable to solve the robot planning problem with innovation and improvement. According to the RoboCup@ home simulation, a new coding structure is redesigned which is called step. This structure has the characteristics of normaliza- tion, complete information and operability. The cost calculation method is used as its fitness function, and change the way of e- volution with ectopic self-crossover and limited mutation. So it can be used for robot mission planning. The effectiveness and su periority of this theory are confirmed by experiments and participating in the competition.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第4期1432-1436,共5页
Computer Engineering and Design
