Aiming at the problems of the RRT*FN algorithm about its slow speed to obtain the path solutions and unable to applied in dynamic environment, a dynamic bidirectional RRT* algorithm with fixed nodes(B-RRT*FND) is proposed, which is used to solve the problem of how to obtain collision-free paths in 2D space quickly in real-time with a robot. The algorithm is based on the RRT*FN algorithm, using the bidirectional greedy search method to speed up the searching and solve the problems of the unidirectional RRT algorithm about its slow searching speed as well as the difficulty of solving in the narrow environment caused by blindly random sampling. Meanwhile, taking the advantage of the fact that fixed nodes do not occupy too much computation in planning, in the process of path iterative optimization, the algorithm updates the map information in real-time, and repairs the broken original path to complete the dynamic path planning. Compared with the RRT, the RRT*FN and other algorithms in three environments, the B-RRT*FND algorithm is superior in planning speed, length of path and dynamic programming performance.