Supplemental Human-Machine Interface for Haptic Shared Control to Convey Sensor Reliability Information to an Operator of Underwater Vehicles.

Abstract

In haptic shared control (HSC) for teleoperation, the human operator and autonomous system share control via torque on the input device. When a discrepancy exists between the human's maneuvering intent and the autonomous maneuver presentation, the human can apply additional force to the device to achieve the desired maneuvering. The required force level is generally predetermined during design and closely relates to the human workload. This study investigated scenarios where discrepancies between human and machine control occur frequently owing to unreliable sensor information. We propose a method to allow the operator to adjust the machine's assist level, facilitating smoother control transitions and proving useful when machine suggestions are unreliable. Conventional HSC systems generally lack sufficient information for operators to make informed decisions regarding adjusting assist levels, leading to delayed decisions and increased workloads. To address this, our research objective is to communicate sensor reliability to humans, thereby facilitating deciding the machine's assist level. We focused on the HSC of a remotely operated underwater vehicle (ROV) where the sensor system is prone to unreliable. Maneuvering experiments with an ROV simulation involving sixteen participants demonstrated the proposed method effectively communicates sensor reliability to operators and reduces mental demand compared to conventional methods.