CLOT: Multi-Robot Motion Planning Via Collaborative Optimal Transport under Signal Temporal Logic Tasks
Ying Zhang, Yunyi Zhang, An Thai Le, Meng Guo
AI summary
Problem
Existing planners struggle to simultaneously satisfy complex, conflicting constraints like collision avoidance, formation, and connectivity while fulfilling high-level temporal tasks, often suffering from poor scalability or getting trapped in local minima.
Approach
The authors introduce CLOT, a gradient-free framework that formulates trajectory optimization as a collaborative optimal transport problem, using parallelized Sinkhorn steps and a hybrid search over robot planning sequences to efficiently navigate nonlinear costs and STL specifications.
Key results
- GPU-accelerated planning completes in seconds for fleets exceeding 100 robots
- Seamlessly enforces complex Signal Temporal Logic tasks including formation and connectivity
- Demonstrates superior feasibility and optimality versus analytical and sampling-based planners
- Validated across diverse simulations and real-world UAV hardware experiments
Why it matters
Provides a scalable, real-time planning foundation for large-scale autonomous fleets operating in safety-critical, dynamic environments.
Abstract
Multi-robot systems often need to navigate clut- tered environments while performing complex tasks. To ensure collision-free trajectories among the robots and with the ob- stacles is essential for the overall safety, along with additional requirements such as dynamic feasibility, relative formation, connectivity maintenance and temporal tasks. Existing work mostly focuses on the design of analytical controllers that encap- sulate all these constraints, which often suffer from undesired local minima due to conflicting non-convex objectives. This work proposes a novel motion planning scheme for multi-robot systems under various safety and high-level tasks, specified as signal temporal logic (STL) formulas over collective states such as collision avoidance, relative formation and connectiv- ity maintenance. A gradient-free method called collaborative optimal transport (CLOT) is proposed that optimizes batches of system-wide smooth trajectories over highly nonlinear costs handled through the zero-order Sinkhorn Step. Via parallel computation on GPUs, the method achieves a planning time of few seconds for small teams and maintains tractability for over 100 robots. Lastly, its applicability is extensively demonstrated both in simulation and hardware, over complex environments and high-level temporal tasks.