Adviser-Actor-Critic: Eliminating Steady-State Error in Reinforcement Learning Control
This addresses precision control challenges in robotics and real-world applications, representing an incremental improvement by integrating feedback control into RL.
The paper tackled the problem of high-precision control in reinforcement learning, which often suffers from steady-state error due to network inaccuracies and poor sample quality, by introducing Adviser-Actor-Critic (AAC) that combines feedback control theory with RL to enhance goal attainment precision, and it outperformed standard RL algorithms in benchmark tests for precision-critical tasks.
High-precision control tasks present substantial challenges for reinforcement learning (RL) algorithms, frequently resulting in suboptimal performance attributed to network approximation inaccuracies and inadequate sample quality.These issues are exacerbated when the task requires the agent to achieve a precise goal state, as is common in robotics and other real-world applications.We introduce Adviser-Actor-Critic (AAC), designed to address the precision control dilemma by combining the precision of feedback control theory with the adaptive learning capability of RL and featuring an Adviser that mentors the actor to refine control actions, thereby enhancing the precision of goal attainment.Finally, through benchmark tests, AAC outperformed standard RL algorithms in precision-critical, goal-conditioned tasks, demonstrating AAC's high precision, reliability, and robustness.Code are available at: https://anonymous.4open.science/r/Adviser-Actor-Critic-8AC5.