Truncated Rectified Flow Policy for Reinforcement Learning with One-Step Sampling
This addresses the problem of modeling multimodal action distributions in reinforcement learning for robotics and control tasks, though it is incremental as it builds on existing flow matching methods.
The paper tackled the challenge of using expressive generative policies in maximum entropy reinforcement learning by proposing the Truncated Rectified Flow Policy, which enables stable training and one-step sampling, outperforming baselines on most MuJoCo benchmarks.
Maximum entropy reinforcement learning (MaxEnt RL) has become a standard framework for sequential decision making, yet its standard Gaussian policy parameterization is inherently unimodal, limiting its ability to model complex multimodal action distributions. This limitation has motivated increasing interest in generative policies based on diffusion and flow matching as more expressive alternatives. However, incorporating such policies into MaxEnt RL is challenging for two main reasons: the likelihood and entropy of continuous-time generative policies are generally intractable, and multi-step sampling introduces both long-horizon backpropagation instability and substantial inference latency. To address these challenges, we propose Truncated Rectified Flow Policy (TRFP), a framework built on a hybrid deterministic-stochastic architecture. This design makes entropy-regularized optimization tractable while supporting stable training and effective one-step sampling through gradient truncation and flow straightening. Empirical results on a toy multigoal environment and 10 MuJoCo benchmarks show that TRFP captures multimodal behavior effectively, outperforms strong baselines on most benchmarks under standard sampling, and remains highly competitive under one-step sampling.