ICRA 2025 - 4 Papers Accepted!

All four of our papers have been accepted for presentation at the International Conference on Robotics and Automation (ICRA 2025), which will be held in-person from 19 to 23 May 2025 in Atlanta, GA, USA.

One of these papers presents a novel, diffusion-based approach top single-agent autonomous exploration, one focuses on a novel, Sheaf-theory-based method to multi-agent pathfinding (MAPF), the third one tackles the multi-robot exploration problem with limited-FOV sensors, while the last one presents a learning-based method for very large-scale MAPF (up to 10,000 agents!) in collaboration with Prof. Jiaoyang Li’s team at CMU.

The preprints of these three papers is available from the publications page and below.

• DARE: Diffusion Policy for Autonomous Robot Exploration
• SIGMA: Sheaf-Informed Geometric Multi-Agent Pathfinding
• MARVEL: Multi-Agent Reinforcement Learning for constrained field-of-View multi-robot Exploration in Large-scale environments
• Deploying Ten Thousand Robots: Scalable Imitation Learning for Lifelong Multi-Agent Path Finding