摘要
现场可编程门阵列(FPGAs)是超大规模可编程专用集成电路,进化算法是能够在线自适应的硬件,它包括进化系统、遗传算法和遗传编程,算法从生物学上求解待定问题的计算方法得到灵感。给出一种基于FPGAs 的新的进化算法,算法中的种群由联想种群和改进种群两个子种群组成且可动态地可重配置,对改进种群中的每个染色体都使用复制、变异和选择操作,不对联想种群而只对改进种群进行变异操作,算法成功地导航机器人在复杂变化的环境中实现避碰。
Field programmable gate arrays (FPGAs) is a programmable very large scale integration (VLSI) circuit. Evolutionary algorithm is hardware which is capable of on-line adaptation, and it includes evolutionary systems, genetic algorithms and genetic pro- gramming. Evolutionary algorithm is a biologically inspired computation method ofproblem solving. A new FPGAs-based evolutionary algorithm is presented which is dynamic and re- configurable. In this algorithm, the population is made up of two sub-populations: memory population andinnovation population. The individuals of the innovation population undergo the operation ofreplication, mutation and selection but the memory population is not mutated. Experimental results show that algorithm proposed can successfully navigate a robot to avoid collision in an unknown or changing environment.
出处
《计算机工程与设计》
CSCD
北大核心
2005年第3期586-587,600,共3页
Computer Engineering and Design
基金
海南省教育厅自然科学基金项目(Hjkj200327)
