Autonomous　behavior　agent　includes　the　robot　most　behaviors　which　are　designed　using　hierarchical　control　method　to　guarantee　their　real　time　performance　for　real　time　navigation　in　response　to　different　situation　perceived.　The　process　of　robot　real　time　navigation　based　on　the　autonomous　behavior　agent　mainly　includes　three　behaviors.　The　sensing　behavior　translates　the　configuration　space　that　the　robot　and　obstacles　exist　in　into　2D　Cartesian　Grid　by　Quadtree　method.　The　path　planning　behavior　designs　the　sub-goals　given　the　global　map,　start　and　goal　points　by　improved　D*　Lite　Algorithm.　And　the　obstacle　avoidance　behaivor　replans　the　path　between　two　adjacent　sub-goals　when　the　environment　changes.　It　is　able　to　replan　faster　than　planning　from　scratch　since　it　modifies　its　previous　search　results　locally　and　enables　robots　adapt　to　the　dynamic　environment.　The　simulation　results　that　are　reported　show　that　the　mobile　robot　navigation　method　is　efficient　and　feasible.