Handling Sparse Rewards in Reinforcement Learning Using Model Predictive Control

Publication Authors M. Dawood; N. Dengler; J. de Heuvel; M. Bennewitz
Published in IEEE International Conference on Robotics & Automation (ICRA)
Year of Publication 2023
Abstract

Reinforcement learning (RL) has recently proven great success in various domains. Yet, the design of the reward function requires detailed domain expertise and tedious fine-tuning to ensure that agents are able to learn the desired behaviour. Using a sparse reward conveniently mitigates these challenges. However, the sparse reward represents a challenge on its own, often resulting in unsuccessful training of the agent. In this paper, we therefore address the sparse reward problem in RL. Our goal is to find an effective alternative to reward shaping, without using costly human demonstrations, that would also be applicable to a wide range of domains. Hence, we propose to use model predictive control~(MPC) as an experience source for training RL agents in sparse reward environments. Without the need for reward shaping, we successfully apply our approach in the field of mobile robot navigation both in simulation and real-world experiments with a Kuboki Turtlebot 2. We furthermore demonstrate great improvement over pure RL algorithms in terms of success rate as well as number of collisions and timeouts. Our experiments show that MPC as an experience source improves the agent's learning process for a given task in the case of sparse rewards.

Type of Publication Conference Proceeding
Lead Image No image
Lead Image Caption
Text
Images
Teaser Image 1
Teaser Image 2 No image
Files and Media
Local Video File
Local PDF File
Settings
Versioning enabled yes
Short name handling-sparse-rewards-in-reinforcment-learning-using-model-predictive-control
Layout
Blocks { "0fc68afc-3b9a-4f14-b574-8cc06b8c9a3d": { "@type": "slate", "value": [ { "type": "p", "children": [ { "text": "" } ] } ], "plaintext": "" } }
Blocks Layout { "items": [ "0fc68afc-3b9a-4f14-b574-8cc06b8c9a3d" ] }
Options
Categorization
Related Items
Contents

There are currently no items in this folder.