Related Experiment Videos
DGR-MAPPO: Enhancing Multi-Agent Cooperative Exploration in Unknown Environments with Distance-Aware Communication
Chufang Wang1, Xiai Chen1, Aoqi Shen1
1School of Mechanical and Electrical Engineering, China Jiliang University, No. 258 Xueyuan Street, Hangzhou 310018, China.
Sensors (Basel, Switzerland)
|August 13, 2026
Summary
This study introduces a novel multi-agent proximal policy optimization (MAPPO) algorithm to enhance collaborative exploration. The new approach improves exploration coverage and efficiency in unknown environments.
Area of Science:
- Robotics
- Artificial Intelligence
- Multi-agent Systems
Background:
- Multi-agent collaborative exploration faces challenges like limited local observations, global collaboration difficulties, and low efficiency.
- Existing methods struggle with dynamic environments and effective information sharing.
Purpose of the Study:
- To develop an advanced multi-agent proximal policy optimization (MAPPO) algorithm for efficient collaborative exploration.
- To enhance exploration coverage, reduce path overlap, and improve adaptability in unknown environments.
Main Methods:
- Implemented a MAPPO algorithm with distance awareness and gated recurrent unit optimization.
- Constructed a composite observation space integrating global maps and local perceptions.
- Introduced a global average pooling layer for stable feature extraction and a distance-based dynamic communication mechanism for information sharing.
Main Results:
- Significantly reduced path overlap rate in untrained complex maps.
- Markedly enhanced exploration coverage and training convergence speed.
- Demonstrated improved collaborative efficiency and adaptability to varying environments.
Conclusions:
- The proposed MAPPO algorithm effectively addresses limitations in multi-agent collaborative exploration.
- The integration of distance awareness and dynamic communication enhances overall system performance and adaptability.
