Paper published to Ocean Engineering by Sara Aldhaheri

Sara Aldhaheri has published her new paper, U-GPMP: A probabilistic inference-based motion planner for underwater mobile manipulators, addressing a key challenge in marine robotics: planning reliable motion for underwater robots operating under uncertainty, hydrodynamic disturbances, and strong vehicle–manipulator coupling.

The work reframes motion planning as a probabilistic inference problem over trajectories, rather than a purely geometric or sampling-based approach. Trajectories are modeled as continuous-time Gaussian processes, and planning is formulated as a maximum a posteriori (MAP) estimation problem on a factor graph.

A central contribution of the study is the introduction of a current-aware Gaussian process prior, which explicitly incorporates ocean current disturbances into the motion model. This enables underwater robots to anticipate environmental effects in advance, resulting in smoother, safer, and more efficient trajectories for intervention tasks.

The research was led by Sara, who carried out the work from theoretical development through to full system-level validation on an underwater vehicle–manipulator platform.