Data-Driven Actuator Model-Based Teleoperation Assistance System

In the construction industry, teleoperation plays a vital role in enabling operators to control machinery from a safe distance. However, this approach requires operators to undergo significant training to become proficient in operating the machines, as they have multiple degrees of freedom and are remotely controlled on a joint level using individual joysticks or levers. Additionally, operators can only confirm the machine's resulting motion after the actual execution, making it challenging to pre-plan the motion of the teleoperated hydraulic machine. To address this problem, a data-driven approach is employed in this study to capture the hydraulic system's nonlinear dynamics, allowing for the modeling of the nonlinear relationship between control inputs and system state changes with less effort and without significant hardware modifications. An assistance system is proposed based on this data-driven actuator model, which pre-visualizes the hydraulic machine's motion within the captured 3D workspace. This pre-visualization enables operators to preview the machine's motion for any sequence of control inputs before execution. The proposed framework is implemented on the full-scale construction machine, BROKK 170, and evaluated within the common construction task of inserting a chisel into a borehole.