Publication

Field Testing of a Cooperative Multi-Robot Sample Return Mission in Mars Analogue Environment

Roland Sonsalla, Florian Cordes, Leif Christensen, Thomas M. Roehr, Tobias Stark, Steffen Planthaber, Michael Maurus, Martin Mallwitz, Elsa Andrea Kirchner

In: Proceedings of the 14th Symposium on Advanced Space Technologies in Robotics and Automation (ASTRA 20017). ESA/Estec Symposium on Advanced Space Technologies in Robotics and Automation (ASTRA-2017) 14th June 20-22 Leiden Netherlands ESA 6/2017.

Abstract

Complex robotic missions become increasingly important for the exploration of our solar system. Future missions are calling for more more sophisticated experiments, the retrieval of samples or even the preparation of manned missions to celestial bodies such as the Moon or Mars. To address these challenges, a cooperating robotic team for (semi-) autonomous exploration of planetary surfaces is introduced in contrast to the currently common single system mission approach. This paper presents the evaluation of a heterogeneous robotic team for planetary exploration purposes. An extensive test campaign with a duration of four weeks was conducted in October/November 2016 in the desert of Utah, USA. The employed robotic systems were tested on natural and unstructured Mars analogue terrain and remotely operated from a control station in Bremen, Germany. The full paper details the performed system tests as well as the conducted cooperative mission sequences in the scope of a sample return mission. Furthermore, the planning and preparation of the field trial campaign as well as the infrastructure set-up in Utah and Bremen and the test execution will be presented with regard to lessons learned in the field and at the control center in Bremen. The field trials were performed using multiple heterogeneous robotic systems including mobile rovers as well as immobile elements. These are as mobile elements (i) the 150kg hybrid walking and driving rover SherpaTT with an actively articulated suspension system and (ii) the micro rover Coyote III with high mobility performance in unstructured terrain. The robotic team was completed by modular immobile elements as (iii) a base-camp acting as communication relays and junction point and (iv) modular payload items containing sensors and sample acquisition units. In order to deploy and handle the immobile elements, both rovers are equipped with manipulators and standardized electro mechanical interfaces. Besides a local control station in the field, a new approach for long distance mission conrol from Bremen was used, utilizing an active upper body exosceleton, allowing force-feedback control, within a 3D virtual reality CAVE system. The field trial campaign was based on a three layered system evaluation: (i) single system tests concerning locomotion capabilities and performance parameters, (ii) single system tests concerning autonomy and cooperative tasks and (iii) full system test in a simulated mission scenario with the primary focus on the execution of a semi-autonomous sample return mission sequence, including all robotic systems as listed above. In the demonstrated mission sequence, SherpaTT acted as exploration and sample acquisition rover. Coyote III took the role of a scout and shuttle rover, collecting the sample containers and transporting them back to the sample return stage. The mission command and control was based in a ground control station located in Bremen, Germany, using a satellite link for communication with the Utah test site. Apart from the demonstration and evaluation of a mission sequence, performance parameters of the different robotic systems were tested within the Mars analogue terrain. Furthermore, different aspects of autonomous exploration tasks could be evaluated on single system level, as well as within multi-robot tests. This includes the evaluation of various locomotion parameters, cooperative mapping and autonomous navigation as well as payload exchange between the robots and force feedback control from Bremen. The paper presents the implemented layered test approach to address the verification of a complex heterogeneous system, ranging from the control center operations to the multi-robot cooperation in analogue Martian terrain. In this context the lessons learned over the course of the field trials, including the need for flexible test organization with multiple systems, the infrastructure set-up and how to deal with unforeseen repair challenges are discussed.

Projekte

3A-sonsalla.pdf (pdf, 5 MB)

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz