Asymmetric Information Enhanced Mapping Framework for Multirobot Exploration based on Deep Reinforcement Learning

2024-04-28

AI Summary

To address multi-robot collaborative exploration in unknown environments, this paper proposes AIM-Mapping: an Asymmetric Information-enhanced deep reinforcement learning mapping framework. Methodologically, it introduces (i) a novel asymmetric actor-critic training paradigm leveraging privileged information (e.g., global ground truth) to construct environment representations and generate supervision signals; (ii) a mutual information-based evaluation module to enhance feature discriminability; and (iii) a topology-graph-matching-driven long-horizon goal allocation mechanism, integrated with geometric-distance-based frontier point identification for efficient coordination. Evaluated in the Gibson simulation environment, AIM-Mapping achieves significant improvements in exploration efficiency (+28.6%) and task completion rate (+34.1%), demonstrating its effectiveness, robustness, and scalability in photorealistic scenarios.

Technology Category

• Intelligent Robots: Multi-Robot Systems
• Multiagent Systems: Multiagent Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Despite the great development of multirobot technologies, efficiently and collaboratively exploring an unknown environment is still a big challenge. In this paper, we propose AIM-Mapping, a Asymmetric InforMation Enhanced Mapping framework. The framework fully utilizes the privilege information in the training process to help construct the environment representation as well as the supervised signal in an asymmetric actor-critic training framework. Specifically, privilege information is used to evaluate the exploration performance through an asymmetric feature representation module and a mutual information evaluation module. The decision-making network uses the trained feature encoder to extract structure information from the environment and combines it with a topological map constructed based on geometric distance. Utilizing this kind of topological map representation, we employ topological graph matching to assign corresponding boundary points to each robot as long-term goal points. We conduct experiments in real-world-like scenarios using the Gibson simulation environments. It validates that the proposed method, when compared to existing methods, achieves great performance improvement.

Problem

• Enhances multirobot exploration using asymmetric information mapping
• Improves environment representation through privileged training data
• Optimizes long-term goal assignment via topological graph matching

Innovation

• Asymmetric actor-critic training with privilege information
• Topological map matching for robot goal assignment
• Mutual information evaluation module for exploration performance