Abstract: Aiming at the deficiencies of traditional path planning algorithms based on grid maps,a twofold-interpolation-based path planning algorithm called T^* is presented.First,using FMM(fast marching method) interpolation,a goal-wave map is obtained,on which wave is supposed to start at the goal node and propagate radiantly through four adjacent nodes, assigning monotonically increasing values to nodes.Based on the goal-wave map,a linear interpolation is used to search a smooth path starring from the robot's current position to the target point along the reverse direction of the wave.At last, an improved VVM(virtual vehicle method) is applied to tracking this generated path by a nonholonomic mobile robot.The experiment results demonstrate the effectiveness of the approach.