Executing long-horizon multi-arm tasks online in complex dynamic environments remains a challenge, particularly when safety is a priority. A new paper on arXiv proposes a distributed dynamic Gaussian belief propagation (DD-GBP) method that directly plans joint-space trajectories for multiple robot arms.
The method is designed for anytime planning and continuous execution, meaning it can refine plans as new information arrives while the robots are already moving. By distributing the belief propagation across agents, it aims to handle the computational and coordination demands of multi-arm systems in real time.
Because the source is a single abstract, details on experimental validation or comparisons are not yet available. The contribution is presented as a planning approach that addresses safety and online execution together, rather than as separate concerns.