Abstract: The efficient launch capability of carrier-based aircraft is one of the important indicators to measure the comprehensive combat effectiveness of aircraft carriers. In order to improve the launch rate of carrier-based aircraft, a guidance path planning method based on Deep Deterministic Policy Gradient is proposed. Firstly, the proposed method models the path planning problem of carrier-based aircraft as a sequential decision-making problem, constructing the state space of the aircraft deck environment and the continuous action space for carrier-based aircraft. Then, based on the carrier-based aircraft motion model, turning angle, obstacle avoidance and other multiple constraints, a reward function is designed that takes into account both immediate reward and long-term cumulative return, which improves the convergence speed of reinforcement learning algorithm. Combined with the relative velocity obstacle method and dynamic sampling strategy, the obstacle avoidance learning ability of the algorithm is enhanced. In order to improve the smoothness of the path, the B-spline curve fitting is used to optimize the planned path to meet the needs of real tasks. The results of simulation experiments on Unity3D show that the proposed method is superior to the compared algorithms in terms of convergence speed, path length, smoothness and other evaluation indicators.