Adaptive Robot Coordination (ARC) framework is a hybrid approach for multi-robot motion planning (MRMP), employing local subproblems to resolve inter-robot conflicts. ARC dynamically adjusts the coordination level by exploring both coupled and decoupled planning spaces, demonstrating probabilistic completeness, adaptability for any robot, and efficient, cost-effective solutions within reduced planning times.

ARC Framework

MRMP framework

  • Compute initial paths through sampling and querying individual roadmaps
  • Detect conflicts among two or more robots
  • Define subproblems around conflicts to obtain feasible paths

Conflict Resolution through Subproblems

  • Upon conflict detection, subproblems are defined around conflicts.
  • Solutions to subproblems are utilized to repair robot paths and resolve conflicts.
  • When subproblems are unsolvable, we adapt them to consider additional robots or space.

Evaluation: ARC was compared against decoupled, coupled, and hybrid baselines in scenarios requiring low, high, and varying levels of coordination.

  • ARC stands out as the sole method capable of addressing the varying coordination scenario, encompassing conflicts with diverse levels of complexity and involving varying numbers of robots.
  • ARC competes effectively with decoupled approaches, which demonstrate scalability in scenarios with low coordination, and coupled approaches, which excel in scenarios demanding higher levels of coordination.

E-ARC: Applying Experience-based planning to the ARC framework

E-ARC is an experience-based approach designed to improve multi-robot motion planning by leveraging previously stored solutions to resolve conflicts efficiently. Traditional experience-based motion planning methods are effective for single-robot problems, but extending them to multi-robot scenarios is challenging due to the exponential growth of the planning space. E-ARC addresses this by identifying and reusing solutions for smaller, relevant subproblems rather than storing complete solutions for full multi-robot instances. This method allows for scalable and efficient motion planning, enabling robots to adapt to complex and dynamic environments.

Evaluation: E-ARC was analyzed in different multi-robot environments to assess its ability to improve planning efficiency and scalability.

E-ARC’s approach was evaluated in different multi-robot systems, including mobile and manipulator robots, demonstrating significant improvements in motion planning.

  • Effectively scales to complex multi-robot scenarios by focusing on relevant subproblems.
  • Reduces computational overhead by leveraging precomputed solutions for common interactions.

K-ARC: Kinodynamic planning using the ARC framework

TODO

Publications

No papers listed in our current papers database.

Updated: