Interaction detection for underwater manipulation with small-scale robots under the influence of currents

Abstract Performing dexterous manipulation underwater with small-scale robots is challenging due to unpredictable current disturbances and the impracticality of integrating force/torque sensors. We introduce a probabilistic interaction detection method based on a Gaussian mixture model that treats wrenches resulting from steady currents as a quasi-static background and interaction wrenches as a dynamic foreground, relying solely on onboard IMU and DVL measurements. In controlled basin experiments with a BlueROV2 platform lifting known payloads under no, weak, medium, and strong currents, our approach reliably detected interaction forces down to 8 N, achieving up to 95 % true positive rate (TPR) and below 2 % false positive rate (FPR) for currents up to 10 N, with performance degradation noted only for very low payloads (4 N) and strong currents (20 N). This method advances autonomous underwater operations by enabling accurate, safe, and adaptive force estimation for precise manipulation in current-influenced environments.