Inspired by the negative effects of the asymptotically optimal rapidly-exploring random tree (RRT*) applied to the robot path planning,such as the low accuracy and worse environmental adaptability,an improvement of RRT* algorithm based on sparse nodes and bidirectional interpolation was presented.RRT* algorithm was carried out with the application of both target bias sampling and sparse node methods.The initial path searching efficiency was enhanced considerably by avoiding excessive searching of the local region.In addition,the redundant nodes in the initial path were eliminated using the principle of triangular inequality.By comparison,the sub-optimal path was achieved in a shortened time caused by the optimized path nodes under the approach of the bidirectional interpolation.Experimental results with a variety of simulation conditions reveal that compared with RRT* algorithm,Informed-RRT* algorithm,and Q-RRT* algorithm,for the initial path planning efficiency,up to 61% enhancement is observed with the proposed algorithm.Meanwhile,the sub-optimal path planning efficiency is increased by 59%,exhibiting an excellent stability with a variety of environments.Finally,the effectiveness of the proposed algorithm is verified experimentally in the tests of practical robot path planning.