This　paper　presents　a　new　method　for　developmental　robot　based　on　a　learning　automaton.　This　method　can　be　considered　as　active　learning　to　select　the　best　action　in　order　to　quickly　adapt　environment.　The　new　model　built　in　the　framework　of　learning　automata　theory　is　an　abstract　and　formal　mathematical　tool　to　describe　the　cognitive　behavior　or　cognitive　development　mechanisms　and　provide　an　effective　logical　structure　for　the　design　of　cognitive　and　development　robots.　The　model　reflects　the　design　principles　of　cognitive　development　robotics.　The　direction　of　cognition　and　development　is　towards　entropy　minimization　and　simulation　results　verify　its　effectiveness.