Policy gradient methods in actor-critic reinforcement learning (RL) have become perhaps the most promising approaches to solving continuous optimal control problems. However, the trial-and-error nature of RL and the inherent randomness associated with solution approximations cause variations in the learned optimal values and policies. This has significantly hindered their successful deployment in real life applications where control responses need to meet dynamic performance criteria deterministically. Here we propose a novel phased actor in actor-critic (PAAC) method, aiming at improving policy gradient estimation and thus the quality of the control policy. Specifically, PAAC accounts for both $Q$ value and TD error in its actor update. We prove qualitative properties of PAAC for learning convergence of the value and policy, solution optimality, and stability of system dynamics. Additionally, we show variance reduction in policy gradient estimation. PAAC performance is systematically and quantitatively evaluated in this study using DeepMind Control Suite (DMC). Results show that PAAC leads to significant performance improvement measured by total cost, learning variance, robustness, learning speed and success rate. As PAAC can be piggybacked onto general policy gradient learning frameworks, we select well-known methods such as direct heuristic dynamic programming (dHDP), deep deterministic policy gradient (DDPG) and their variants to demonstrate the effectiveness of PAAC. Consequently we provide a unified view on these related policy gradient algorithms.
We address the issue of estimation bias in deep reinforcement learning (DRL) by introducing solution mechanisms that include a new, twin TD-regularized actor-critic (TDR) method. It aims at reducing both over and under-estimation errors. With TDR and by combining good DRL improvements, such as distributional learning and long N-step surrogate stage reward (LNSS) method, we show that our new TDR-based actor-critic learning has enabled DRL methods to outperform their respective baselines in challenging environments in DeepMind Control Suite. Furthermore, they elevate TD3 and SAC respectively to a level of performance comparable to that of D4PG (the current SOTA), and they also improve the performance of D4PG to a new SOTA level measured by mean reward, convergence speed, learning success rate, and learning variance.
High variances in reinforcement learning have shown impeding successful convergence and hurting task performance. As reward signal plays an important role in learning behavior, multi-step methods have been considered to mitigate the problem, and are believed to be more effective than single step methods. However, there is a lack of comprehensive and systematic study on this important aspect to demonstrate the effectiveness of multi-step methods in solving highly complex continuous control problems. In this study, we introduce a new long $N$-step surrogate stage (LNSS) reward approach to effectively account for complex environment dynamics while previous methods are usually feasible for limited number of steps. The LNSS method is simple, low computational cost, and applicable to value based or policy gradient reinforcement learning. We systematically evaluate LNSS in OpenAI Gym and DeepMind Control Suite to address some complex benchmark environments that have been challenging to obtain good results by DRL in general. We demonstrate performance improvement in terms of total reward, convergence speed, and coefficient of variation (CV) by LNSS. We also provide analytical insights on how LNSS exponentially reduces the upper bound on the variances of Q value from a respective single step method