[On-going project] Concept:
Technology to create a digital twin of an actual robot using external sensors built with ETR Lab's own technology.
In this study, we verified the main inverse kinematics (IK) solutions using MuJoCo simulation and evaluated the agreement with the actual robot. After obtaining the IK solution based on the DH (Denavit-Hartenberg) parameters of the robot, the path generated by MuJoCo was identically reproduced on the actual robot, and the robot's pose was measured using an infrared position tracker. In particular, the trajectory was generated by interpolating the passing points during the pose transition between various standard poses (vertical upright, horizontal full stretch, 90-degree bending of the upper arm vertically and forearm horizontally, etc.). In MuJoCo, the passing points generated through the computation were continuously tracked, while in the actual robot, the robot's motion was stopped at certain time intervals and measurements were performed for accurate measurements when transitioning between standard poses. Through this, the distance difference between the trajectory generated by MuJoCo and the trajectory of the actual robot was analyzed, and the error rate was quantitatively evaluated. The analysis results show that the IK solver used in this study has high accuracy and can be applied to various robot systems assuming that no linear motion joints are included. This study can contribute to securing the reliability of the IK solution in the robot control and simulation-based verification process, and can be used as basic data for additional verification and optimization studies in various environments in the future.
Applications: External manipulation via VR and PCs, enabling precise control. Practical robot and model synchronization for extreme testing, including collision tests and high-performance evaluations.
[On-going project] Concept:
Technology to create a digital twin of an actual robot using external sensors built with ETR Lab's own technology.
In this study, we verified the main inverse kinematics (IK) solutions using MuJoCo simulation and evaluated the agreement with the actual robot. After obtaining the IK solution based on the DH (Denavit-Hartenberg) parameters of the robot, the path generated by MuJoCo was identically reproduced on the actual robot, and the robot's pose was measured using an infrared position tracker. In particular, the trajectory was generated by interpolating the passing points during the pose transition between various standard poses (vertical upright, horizontal full stretch, 90-degree bending of the upper arm vertically and forearm horizontally, etc.). In MuJoCo, the passing points generated through the computation were continuously tracked, while in the actual robot, the robot's motion was stopped at certain time intervals and measurements were performed for accurate measurements when transitioning between standard poses. Through this, the distance difference between the trajectory generated by MuJoCo and the trajectory of the actual robot was analyzed, and the error rate was quantitatively evaluated. The analysis results show that the IK solver used in this study has high accuracy and can be applied to various robot systems assuming that no linear motion joints are included. This study can contribute to securing the reliability of the IK solution in the robot control and simulation-based verification process, and can be used as basic data for additional verification and optimization studies in various environments in the future.
Applications: External manipulation via VR and PCs, enabling precise control. Practical robot and model synchronization for extreme testing, including collision tests and high-performance evaluations.