This paper presents a new methodology where machine learning is used for detecting various levels of slip in the context of planetary exploration robotic missions. This methodology aims at employing proprioceptive rover sensor signals. Consequently, no operational complexity is added to the rover's commanding and it is independent of lighting conditions. Two supervised learning methods (Support Vector Machines and Artificial Neural Networks) are compared to two unsupervised learning approaches (K‐means and Self‐Organizing Maps (SOM)). Physical experiments using a single‐wheel testbed equipped with an MSL spare wheel and a real planetary exploration rover validate the implemented methodology. Performance is evaluated in terms of well‐known metrics both considering single data points and subsets of consecutive data points (moving median filter). Computation time and storage requirements are also examined. One of the SOM‐based algorithms, semantic SOM method, demonstrates a proper balance between the benefits of supervised learning algorithms (high success rate, >96%) and the advantages of unsupervised learning methods (low storage requirements, 5 kb, and no need of manually‐labeled training data). This paper also addresses the most convenient placement of IMU sensors on the rover chassis such that slippage detection is maximized.
Robotic rappelling is an intriguing concept for exploration of planetary craters and their Earth analogs, volcanoes.Integrating a tensioned tether to a framewalking robot enables a new statically-stable locomotive capability appropriate for rappelling on steep and rugged terrains. Rappelling with a tether-assisted framewalker also allows eficient execution of multi-level control. These ideas are manifested in the locomotion configuration of Dante II. The appropriateness of the Dante I1 configuration for rappelling was evaluated during a variety of tests and its 1994 exploration of the active volcano of Mount Spurr in Alaska.
This paper shows the performance of various traction control strategies that aim to minimize slippage and wheel fighting by properly adjusting the velocity of each traction wheel in a planetary rover. These strategies are validated through simulations performed in AN-VEL (Quantum Signal LLC) and using two rovers currently employed by NASA. These experiments use similar features to those that a planetary rover would face on the Moon such as terrain geomorphology and lunar gravity. After running those experiments, the following conclusions were drawn: (1) when no traction control is considered, results show the rover gets entrapped or makes a shorter progress than when traction control is applied; (2) the proposed traction controllers demonstrate a proper balance between slipcompensation (lowest mean slip) and reduction of wheel fighting effects (less aggressive control actions); (3) after considering two different planetary rovers, it is observed that the mechanical configuration effects slip reduction. These contributions can also be observed in the accompanying videos.
Mobile robotics has seen a wide variety of mechanisms and strategies for motion in diverse terrain. Some robots employ rolling, some use legs for walking, some can hop, and some are capable of multiple of these modes. In this paper, we present the latest Robotic All-Terrain Surveyor (RATS) prototype as a unique design that can emulate a variety of locomotion modes by virtue of its geometric design and type of actuation. The novel robot has a spherical body the size of a soccer ball with 12 legs symmetrically distributed around its surface. Each leg is a single-DOF pneumatic linear actuator, oriented normal to the spherical body. Thorough investigation of this prototype’s mobility and actuation behavior has demonstrated the feasibility of tipping, hopping, and prolonged rolling locomotion by altering the actuation patterns of its legs. Here we summarize the experimental results of this characterization and present an understanding of the system’s performance limitations in an effort to draw insight for controlling its movements. We also discuss the effectiveness of RATS mobility strategies for varied terrains in light of initial testing on flat surfaces.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.