ESA GNC Conference Papers Repository

Technologies for control and cooperation of multi-robot autonomous system in defence and space applications
M.D. Detratti
Presented at:
Salzburg 2017
Full paper:

Team or swarm of robots forming a network of sensors interacting with one other will allow in the foreseeable future to perform complex functions with a higher robustness and efficiency than the single robot systems which are already nowadays replacing humans in several tasks. Applications of multi-robot systems range from smart highways to automated industrial and agricultural environment, from search and rescue to military missions in all the domains supporting multiple capabilities in hazardous environments. (e.g. de-mining and surveillance) Several studies have been and are being carried out within the European Defence Agency (EDA) framework with focus on potential military applications and technology gaps in this field which will represent more than an evolution for the military doctrine. However, the dull, dirty and dangerous tasks and specific defence operational requirements which may justify the use of unmanned multi-robot system for the defence sector, present clear synergies and common challenges with potential space applications. The execution of space mission with multi-robot systems is already addressed in several studies and subject of several research initiatives. Multiple micro-or nano-satellites flying in formation may expand functionally, enhance mission performance and reduce operational cost, while increasing the overall system fault tolerance. Distributed multi-robot applications for the space domain might furthermore include on-orbit operations, planetary exploration and construction. Both in defence and space application domains, tasks have to be often accomplished in unknown or unstructured environments with no established communication network or global navigation and localization systems. Developing an underwater defence capabilities system or an energy scavenging system in Mars it's going to be very difficult and dangerous and better addressed by unmanned systems to eliminate or reduce risks for humans. Permanent surveillance over a wide and remote area and extra-terrestrial bodies exploration can be similarly carried out by a team of autonomous robots with a common coal to a series of targets/points of interest, and gain as much intelligence/knowledge as possible on a team-wide level. Space, as the battlefield, is a harsh and dangerous place, indeed space debris elimination is not that far away from de-mining. Multi-robot teams or swarms represent the 'natural' evolution for such complex and distributed applications with a non-negligible risk of damaging (or losing control of) operational units. Ideally they should able to act autonomously without the need of any infrastructure or any form of external coordination. These systems should be able dynamically network (connect, share, and collaborate) and adapt to the operational condition thus help in reducing the burden on operator control ultimately being able to perform coordinated tasks without support or external intervention. Reliance on multiple agent systems increase the robustness as losing one (or more) element out of the team due to malfunction or external factor (a mortar or an asteroid impact) would minimize the impact on the mission failure. This paper will firstly present an overview of the EDA activities related to autonomous system and multi-robot cooperation research aimed at improving defence capability which may also benefit space applications. Then, some of the technology gaps in the fields of autonomous and cooperative control and failure tolerant approaches and recovery methods will be presented. Even if substantial advancements are being made, additional research is still needed on software, algorithms and hardware before being able to achieve the fully autonomous level as well as to match the size, weight, power and cost performance to the usage in practical implementations. Relevant areas of research in the field of guidance and control which would need more development include (not limited to): optimal information sharing, robust formation control, autonomous decision making and adaptive behaviour, visual servoing and autonomous all-terrain capabilities. The paper will conclude by identifying the still existing gaps to be filled to enable multi-robot teams to robustly, efficiently and safely perform some of the identified applications without (or with limited) human control, and by proposing recommendations for areas of dual-use R&D and innovation taking into account the shortfalls to meet operational needs.