Education Department of Guangxi Zhuang Autonomous Region, Key Laboratory of Advanced Manufacturing and Automation Technology, Guilin University of Technology, Guilin, People’s Republic of China
Auto-generated profile, based on publicly available information
Hongyuan Chen
Education Department of Guangxi Zhuang Autonomous Region, Key Laboratory of Advanced Manufacturing and Automation Technology, Guilin University of Technology, Guilin, People’s Republic of China
To address the issues of exploration imbalance and slow convergence speed in the Q‐learning path planning algorithm, an adaptive improved Q‐learning path planning algorithm based on an obstacle learning matrix and artificial potential field (APF) is proposed. First, an obstacle learning matrix is established to store the positions of obstacles encountered in each learning iteration, avoiding redundant learning of the same obstacle. Second, an adaptive exploration enhancement strategy is introduced by incorporating the concept of success rate. This strategy divides the decay process of the exploration rate into an exploration‐dominant initial stage and an exploitation‐dominant later stage. Finally, an APF‐weighted action selection strategy is introduced, utilizing the guiding force generated by the APF to encourage the mobile robot to avoid obstacles more efficiently. Simulation results show that the proposed method can effectively reduce the number of iterations and time consumption in the optimization process, resulting in a smoother and more stable planned path.
27 February 2026 in Measurement Science and Technology
To address unstable value estimation and insufficient use of temporal information by the soft Actor–Critic (SAC) algorithm in dynamic environments, we propose an improved SAC-based path-planning method that integrates distributional value learning and long short-term memory (LSTM). First, the critic models the return distribution via quantile regression to reduce the variance of value estimates and improve the stability of policy evaluation. Second, an LSTM encodes historical obstacle observations to enhance prediction of environmental dynamics and decision-making. Through comparative experiments in dynamic scenarios on the Gazebo simulation platform, the proposed method outperforms LSTM-enhanced DDPG, vanilla SAC, distributional SAC, and LSTM-SAC in average path length, average steps, and average planning time. Moreover, real-world experiments further verify the feasibility and robustness of the approach.