Distributed controller keeps warehouse robots safely on track
Safety-Filtered Distributed Koopman-MPC
Robotics
Summary
Robots working together in a shared space need to avoid collisions, but communication problems can make safety checks tricky. The authors propose a two-part method where robots use shared plans to predict motion while independently enforcing hard safety limits using local sensing. This safety filter ensures robots stay clear of obstacles and each other even if communication drops. They tested their method on simulated warehouse robots and found it prevented collisions reliably while reaching all goals.
What this means in practice
- •For warehouse automation teams: Control fleets of warehouse robots safely despite communication dropouts by using the proposed safety filter and Koopman-based predictive control.$Commercial implications: Enables collision-free multi-robot coordination in warehouses, offering a product improvement for automation providers managing fleets under unreliable network conditions.
- •For industrial robot integrators: Deploy safe distributed controllers for robots operating near fixed obstacles using local sensing combined with limited trajectory sharing.
Authors
Shengjun Zhang, Wenhao Li, Zhenxin Lin, Zhenglong Sun
Abstract
Distributed model predictive control (DMPC) often constructs both predictions and collision constraints from neighbor trajectories, so packet loss can remove both. We separate these roles: received trajectories drive Koopman-MPC, while local sensing and shelf geometry define a hard-constrained quadratic program (QP) that projects the applied input. Its radial demand is the least constant acceleration that keeps a supporting-plane clearance nonnegative throughout one zero-order-hold interval. Complementary pair rows recover the coupled demand without exchanging safety decisions. We give an intersample separation theorem under bounded snapshot and directional plant errors, an exact max-min test for simultaneous local feasibility, and a sensing-radius condition for switching interaction graphs. Anticipatory high-order rows may be relaxed for performance, but the finite-hold rows contain no safety slack. Matched eight-robot warehouse simulations use a frozen Koopman model, nonlinear drift, bounded inputs and speed, shelf constraints, a 120 ms control period, and packet dropout. The full controller is collision-free in 20/20 matched trials and reaches 160/160 robot goals; predictive Koopman-MPC without the final projection is collision-free in 1/20 trials. All 38,400 full-method hard-row sets pass the online feasibility test, and every local QP solves. Five-stream fleet sweeps are collision-free and hard-row feasible through 16 robots; the 20-robot boundary fails only after the online margin turns negative, while the reconstructed per-agent critical path remains below the sampling period. Bounded-sensing and differential-drive tests provide additional deployment stress.