Deep reinforcement learning for multi-agent coordination
Problem
Multiple robots working in narrow, confined environments suffer congestion and interference that hinder collective task performance, and coordinating them without explicit communication is hard.
Why it matters
Robot teams that avoid congestion could operate in warehouses, tunnels or disaster sites where coordination failures halt work.
What the AI did
A stigmergic multi-agent deep reinforcement learning framework (S-MADRL) uses virtual pheromones to model local and social interactions, plus curriculum learning, enabling decentralized emergent coordination without explicit communication.
Results
Simulation results show most effective coordination of up to eight agents, with self-organized asymmetric workload distributions reducing congestion
Impact
Simulation-only demonstration of congestion reduction; no physical-robot or real-world deployment shown.
Record
| Venue | Artificial Life and Robotics |
|---|---|
| Published | 2025 |
| Work type | article |
| Authors | Kehinde O. Aina, Sehoon Ha |
| Citations | 8 (OpenAlex, at import time) |
| References | 24 |
| Model or method | S-MADRL (stigmergic multi-agent deep RL with curriculum learning) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1007/s10015-025-01089-z |
| Open access | hybrid |
| Keywords | multi-agent reinforcement learning, stigmergy, virtual pheromones, robot coordination, curriculum learning |
Importance rating
real but narrower advance
Congestion in confined multi-robot settings hinders collective tasks; coordination without communication is a genuine capability, but demonstration is simulation-only with up to eight agents and no stated application scale.
More in Robotics & Autonomy
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | Autonomous robotic laparoscopic surgery for intestinal anastomosis | 2022 | 384 |
7
|
| 2 | Solving Rubik's Cube with a Robot Hand | 2019 | 569 |
6
|
| 3 | PaLM-E: An Embodied Multimodal Language Model | 2023 | 346 |
6
|
| 4 | Assistive Navigation Using Deep Reinforcement Learning Guiding Robot With UWB/Voice Beacons and Semantic Feedbacks for Blind and Visually Impaired People | 2021 | 48 |
6
|
| 5 | Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges | 2022 | 154 |
6
|