Collision-free　motion　planning　and　trajectory　tracking　of　underactuated　robots　are　investigated　in　this　paper.　The　partly　stable　planners　are　introduced　and　the　control　objectives　are　fulfilled　by　the　proper　switching　of　partly　stable　planners.　The　genetic　algorithm　is　introduced　in　the　best　switching　sequence　searching.　Penalty　method　is　utilized　and　the　constrained　optimizations　turns　to　be　unconstrained　ones,　then　the　best　solutions　are　obtained　by　the　genetic　algorithm.　At　last　numerical　simulations　which　are　carried　out　on　the　planar　3R　underactuated　robot　show　the　effectiveness　of　the　proposed　method.