ROS 2-Integrated Low-Latency Remote Force Feedback System Using FPGA-Accelerated FOC Control and Inverse Kinematics

In scenarios where precise remote manipulation is required, a highly accurate and low-latency force feedback robotic arm is important for improving the efficiency of manipulation. Because, with the increasing delay of the system, the time required for the robotic arm to complete the task will be extended accordingly, which will make the path error of the robotic arm bigger and bigger, which will seriously affect the maneuvering accuracy and work efficiency. To solve this problem, this study proposes an acceleration control method based on a force-feedback robotic arm. The Field-Oriented Control (FOC) algorithm and the inverse kinematics computation of the robotic arm are accelerated by utilizing the parallel computation capability of field programmable gate arrays (FPGAs), which reduces the system latency while supporting high-precision operation. The FOC algorithm calculates the motor torque based on the measured motor currents and maps this torque to the remotely operated device to achieve force feedback. The inverse kinematics computes the 3D coordinates of the user's hand movements as the end position of the robotic arm, enabling motion tracking. To integrate this complex telecontrol system involving communication and simulation, Robot Operating System 2 (ROS 2) and Gazebo are used for efficient development and real-time simulation with a digital twin model. The average maximum error of force feedback has a Just Noticeable Difference (JND) of 9.54%. The experimental evaluation of the parallel processing accelerated by the FPGA platform shows that the inverse kinematics computation and FOC algorithm computation are completed in only 0.29 ms under system acceleration. The computation from the motor current measurement and target coordinate input to the motor control signal output realizes the low-latency and high-efficiency control of the robotic arm. The remote operating system based on ROS 2 realizes a control cycle period of 6.1 ms.