Development of Control Strategies for a Torque-Controlled Manipulator in a Lunar-Analog Greenhouse

Abstract

The automation of plant cultivation in lunar greenhouses is critical for reducing astronaut workload and ensuring mission sustainability. This thesis presents the design and validation of an impedance-based trajectory tracking framework for the EDEN Versatile End-effector (EVE), a 7-Degree of Freedom (DOF) torque-controlled robotic manipulator. While agricultural robotics often relies on position control, this work addresses the necessity for contact safety in unstructured environments through torque-based strategies and fills a reported research gap by designing and validating impedance-based trajectory tracking controllers tailored for the EVE system. The following four control architectures are implemented and evaluated within a high-fidelity MuJoCo simulation: Proportional-Derivative (PD) controller with gravity compensation, Augmented Proportional-Derivative (PD+) controller, Slotine-Li Controller (SLC), and the Generalized Tracking Controller (GTC). Benchmarking is conducted using trajectories generated via a Quadratic Spline-via interpolator to reflect realistic operational constraints. Results indicate that while the SLC and the two stiffest GTCs, using compliant factors of -0.4 and -0.7, provide better tracking performance and superior robustness to mass and inertia uncertainties of +-10% and center of mass uncertainty of +-5%, the PD+ and the most compliant GTC, using a compliance factor of -0.95 provide better compliance inferring better suitability for physical interaction. In the latter selection, the PD+ architecture offers the most immediate compliant response to external disturbances and a significantly faster convergence rate to the reference trajectory following disturbance rejection, making it the preferred candidate for the 7-DOF torque-controlled robotic manipulator. Furthermore, simulations in lunar gravity (g ≈ 1.62 m/s2) demonstrate a reduction in static torque requirements by approximately 83%, confirming the system’s suitability for lunar deployment and its enhanced flexibility in terms of payloads and acceleration profiles in such low gravity environment.