← Back
ICRA 2026
Manipulator-Effort-Aware MPC for Body Motion Coordination of an Underwater Walking Robot under Ocean Current Disturbances
Bonhak Koo, Bong Huan Jun, Daegil Park
AI summary
An effort-aware MPC controller adaptively adjusts an underwater walking robot's body posture to keep manipulator torques within safe limits during rope-grasping in ocean currents.
Problem
Conventional decoupled control strategies fail to adapt an underwater walking robot's body motion to manipulator burden, leading to instability and torque saturation under ocean-current disturbances during grasping tasks.
Approach
The method integrates manipulator joint torque norms into a Model Predictive Control framework to dynamically coordinate body approach and pitch compensation with arm operations.
Key results
- Simulated rope-grasping under increasing ocean currents (0.5–2.5 knots)
- Baseline decoupled control caused torques to exceed saturation limits above 2.0 knots
- Proposed MPC maintained torques within safe bounds via adaptive body compensation
- Demonstrated improved disturbance rejection and mechanically stable loco-manipulation
Why it matters
Enables safer and more reliable underwater manipulation in dynamic currents, directly benefiting mine clearance and deep-sea inspection missions.
Abstract
No abstract on file.