For　the　search　and　rescue　robot　navigation　in　unknown　environment,　a　bionic　self-learning　algorithm　based　on　A*　and　Q-Learning　is　put　forward.　The　algorithm　utilizes　the　Growing　Self-organizing　Map　(GSOM)　to　build　the　environment　topology　cognitive　map.　The　heuristic　search　A*　algorithm　is　used　the　global　path　planning.　When　the　local　environment　changes,　Q-Learning　is　used　the　local　path　planning.　Thereby　the　robot　can　obtain　the　self-learning　skill　by　studying　and　training　like　human　or　animal,　and　looks　for　a　free　path　from　the　initial　state　to　the　target　state　in　unknown　environment.　The　theory　proves　the　validity　of　the　method.　The　simulation　result　shows　the　robot　obtains　the　navigation　capability.　©　2012　Universitas　Ahmad　Dahlan.