Field Assessment of Force Torque Sensors for Planetary Rover Navigation

TL;DR

This work assesses the potential of force-torque sensors (FTS) for improving planetary rover navigation, focusing on terrain classification and drawbar pull estimation using field data from the MaRTA rover with six FTS and an onboard IMU. Through 1-second window features, PCA/TSNE visualization, and comparisons of SVM and neural network classifiers, the study finds that FTS can enhance terrain classification, particularly for neural networks, and provides initial methods to filter FTS signals for pull estimation. The results, complemented by open BASEPROD data, offer practical guidance on sensor mounting, data fusion with IMU, and the design of future rover upgrades and control algorithms. While direct drawbar pull extraction remains challenging in long traverses, the work outlines a geometrically grounded approach and emphasizes the need for ground-truth validation and controlled experimentation to realize robust traction estimation in planetary environments.

Abstract

Proprioceptive sensors on planetary rovers serve for state estimation and for understanding terrain and locomotion performance. While inertial measurement units (IMUs) are widely used to this effect, force-torque sensors are less explored for planetary navigation despite their potential to directly measure interaction forces and provide insights into traction performance. This paper presents an evaluation of the performance and use cases of force-torque sensors based on data collected from a six-wheeled rover during tests over varying terrains, speeds, and slopes. We discuss challenges, such as sensor signal reliability and terrain response accuracy, and identify opportunities regarding the use of these sensors. The data is openly accessible and includes force-torque measurements from each of the six-wheel assemblies as well as IMU data from within the rover chassis. This paper aims to inform the design of future studies and rover upgrades, particularly in sensor integration and control algorithms, to improve navigation capabilities.

Figures (15)

• Figure 1: ESA's rover testbed MaRTA during the collection of the analogue planetary exploration dataset BASEPROD in Bardenas Reales, Spain Baseprod. The FTS can be seen in each of the six legs with a close-up highlighting the sensor position relative to the wheel axle.
• Figure 2: Rover view of the example terrains classified as (\ref{['fig:terrain-loose']}) "loose", (\ref{['fig:terrain-compressed']}) "compressed", (\ref{['fig:terrain-pebbles']}) "pebbles", and (\ref{['fig:terrain-rock']}) "rock". The images were taken with the rover's front-facing RGB-D camera.
• Figure 3: Simplified kinematics of MaRTA's locomotion platform. In this figure, MaRTA is headed toward the viewer. The black lines indicate the three passive bogies with the thick black lines representing the passive bogie joints. The deployment actuators are indicated in blue, steering in red. The green boxes represent the FTS and the cylinders the wheels. The IMU and its axes can be seen in the rover's center.
• Figure 4: FTS and IMU data on pebbles. The first row shows front right FTS's forces along its three axes, the second row shows the same FTS's torques, and the third row visualizes the IMU's acceleration data. The three columns correspond to the sensors' $x$, $y$, and $z$ axes respectively.
• Figure 5: FTS and IMU data on pebbles. The first row shows front right FTS's forces along its three axes, the second row shows the same FTS's torques, and the third row visualizes the IMU's acceleration data. The three columns correspond to the sensors' $x$, $y$, and $z$ axes respectively.