# Theoretically Guaranteed Policy Improvement Distilled from Model-Based Planning

arXiv: https://arxiv.org/abs/2307.12933 · HTML: https://mlsanity.com/paper/2307.12933/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 5 of its 28 references that are in the map.
- First posted on arXiv 24 Jul 2023.

## Builds on (most cited first, 5 of 5)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Control](https://mlsanity.com/paper/1812.00568/) (2018)
- [Model-Based Value Estimation for Efficient Model-Free Reinforcement Learning](https://mlsanity.com/paper/1803.00101/) (2018)
- [Accelerating Reinforcement Learning for Autonomous Driving using Task-Agnostic and Ego-Centric Motion Skills](https://mlsanity.com/paper/2209.12072/) (2022)

## Abstract (arXiv)

Model-based reinforcement learning (RL) has demonstrated remarkable successes on a range of continuous control tasks due to its high sample efficiency. To save the computation cost of conducting planning online, recent practices tend to distill optimized action sequences into an RL policy during the training phase. Although the distillation can incorporate both the foresight of planning and the exploration ability of RL policies, the theoretical understanding of these methods is yet unclear. In this paper, we extend the policy improvement step of Soft Actor-Critic (SAC) by developing an approach to distill from model-based planning to the policy. We then demonstrate that such an approach of policy improvement has a theoretical guarantee of monotonic improvement and convergence to the maximum value defined in SAC. We discuss effective design choices and implement our theory as a practical algorithm -- Model-based Planning Distilled to Policy (MPDP) -- that updates the policy jointly over multiple future time steps. Extensive experiments show that MPDP achieves better sample efficiency and asymptotic performance than both model-free and model-based planning algorithms on six continuous control benchmark tasks in MuJoCo.