Modeling, Simulation and Control of a Robotic Arm for Planetary Exploration

Kurzfassung

The next milestone in Mars exploration will be a sample return to use state-of-the-art laboratories on Earth to analyze the Mars soil in-depth. With the NASA-ESA Mars Sample Return campaign, this subsequent chapter in Mars exploration is addressed. One part ESA is contributing to this campaign is its Sample Fetch Rover, whose task will be to collect different soil samples, already prepared and stored in small tubes at various areas inside the Jezero Crater by the Perseverance rover, and carry them to NASA’s Mars Ascent Vehicle to be launched into Mars orbit and then transported back to Earth. This study aimed for the development of a model for the control of ESA’s Sample Fetch Rover robotic arm to simulate its kinematic and dynamic behavior using the KUKA KR 10 R1100-2 robot manipulator predetermined by the company on whose cooperation this master thesis was based. Therefore, first, a literature review in space robotics had been conducted. Followed by an in-depth analysis of various control strategies resulting in an inverse dynamics control selected as a suitable control strategy for this study. Additionally, through an analysis of different available modeling software the Spacecraft Robotics Toolkit was selected and used to create the robot model with certain simplifications in its geometry and mass distribution. Furthermore, the Robotics System Toolbox was used to generate two example trajectories simulating the manipulator picking up a Mars sample tube and storing it into the rover’s backpack. The next step was the realization of the Simulink model, based on the previously selected inverse dynamics control strategy. Therefore, a control loop with various subsystems considering the input trajectories and the manipulator’s forward and inverse kinematics and dynamics, was developed and tuned to accurately simulate the manipulator’s motion. To analyze the accuracy of the Simulink model, each simulated position orientation, and linear and angular velocity parameter of the end-effector was plotted against their respective desired input parameter. An error analysis led to a maximum position error of 5.35 mm with a motion stabilization after approximately four seconds and a maximum orientation deviation of 3.9°. Finally, various recommendations were presented to enhance the current model for a more realistic simulation of the manipulator’s behavior.