Actuator Blockage Detection in Milligravity

Abstract

Exploration rover like IDEFIX, the rover of the Martian Moons Exploration Mission (MMX), even when designed with extraordinary diligence, can get stuck in their environment or be affected by other faults. Ground loops for fault detection and isolation are time consuming, especially in a time-limited mission like that of the MMX rover. Thus, on-board fault detection, isolation and recovery (FDIR) in case a rover gets blocked with their actuators is very advantageous. The MMX rover is provided with four identical shoulder modules including a leg and a wheel with one motor each. The available position, torque and current sensors shall be used to detect a blockage and classify if it is positioned at a leg or a wheel to induce the corresponding recovery strategy. To analyze this problem, tests on flight-like hardware were conducted, whereby blockages were induced at different locations at the leg and wheel. In a first analysis, it was observed that only current and torque signals can be used to detect a blockage, but none of the signals can be used alone to isolate a blockage. With the gathered data, logistic regression was used to classify the blockage location in almost all cases correctly. With this result, blockages can be detected on-board and recovered in a further step, which increases the autonomy and hence the reliability of the MMX rover locomotion.