In　order　to　realize　the　self-learning　of　robot　trajectory　tracking,　this　paper　presents　an　automata　model　based　on　Skinner＇s　conditioning　operation　(SKCOA).　The　model　consists　of　states　set,　actions　set,　states　transition　mechanism,　probabilities　set,　orientation　function,　learning　mechanism　and　entropy　of　operating　behavior,　etc.　In　the　simulation　experiment,　robot　learns　3　different　trajectories　based　on　SKCOA　automata　and　test　it　by　tracking　linear　and　circular　trajectories.　The　experimental　results　show　that　the　robot　has　a　certain　ability　to　learn,　and　can　track　the　trajectory　well.