DR-MPC: Deep Residual Model Predictive Control for Real-world Social Navigation

TL;DR

This paper introduces DR-MPC, a hybrid approach combining model predictive control and deep reinforcement learning, enabling robots to navigate safely in crowded environments with minimal training data and improved performance over prior methods.

Contribution

The paper presents DR-MPC, a novel method that integrates MPC with DRL to enhance real-world social navigation for robots, reducing data needs and increasing safety.

Findings

DR-MPC outperforms prior DRL methods in simulation.

Robots navigate crowded spaces with fewer errors.

Successful real-world experiments with limited training data.

Abstract

How can a robot safely navigate around people with complex motion patterns? Deep Reinforcement Learning (DRL) in simulation holds some promise, but much prior work relies on simulators that fail to capture the nuances of real human motion. Thus, we propose Deep Residual Model Predictive Control (DR-MPC) to enable robots to quickly and safely perform DRL from real-world crowd navigation data. By blending MPC with model-free DRL, DR-MPC overcomes the DRL challenges of large data requirements and unsafe initial behavior. DR-MPC is initialized with MPC-based path tracking, and gradually learns to interact more effectively with humans. To further accelerate learning, a safety component estimates out-of-distribution states to guide the robot away from likely collisions. In simulation, we show that DR-MPC substantially outperforms prior work, including traditional DRL and residual DRL models. Hardware experiments show our approach successfully enables a robot to navigate a variety of crowded situations with few errors using less than 4 hours of training data.