In order to solve the problems of traditional ant colony algorithm in mobile robot path planning, such as low convergence speed, low quality of convergence path, deadlock, and poor dynamic obstacle avoidance capability, a path planning method based on improved obstacle avoidance strategy and double optimization ant colony algorithm (DOACO) was proposed. Firstly, a probability transfer method was designed. The pseudo-random probability adjustment factor was introduced to adjust the selection degree of high-quality path points in the probability transfer function. It avoided the problem that the probability of selecting high-quality path points in traditional ant colony algorithm was too low. Secondly, the weight of each component of the probability transfer function was adaptively adjusted to optimize the convergence speed of the algorithm. Then the elite saving strategy was introduced to prevent the data falling back of the algorithm. The elite saving strategy can also improve the quality of the path. In order to further improve the quality of path, a path optimization strategy was proposed based on key path points. This strategy tried to generate better path segments by looking for key path points. Finally, a obstacle avoidance strategy based on obstacle avoidance behavior and local path replanning was proposed to solve the problems of poor obstacle avoidance ability and lack of real-time performance. The experimental results showed that compared with the traditional ant colony algorithm, DOACO algorithm can not only plan a better path, but also can have a faster convergence speed, and the obstacle avoidance strategy can effectively deal with a variety of collision situations.