Blueswarm: 3D Self-organization in a Fish-inspired Robot Swarm
MetadataShow full item record
CitationBerlinger, Florian. 2021. Blueswarm: 3D Self-organization in a Fish-inspired Robot Swarm. Doctoral dissertation, Harvard University Graduate School of Arts and Sciences.
AbstractAnimals team up to collectively address challenges they could not overcome individually. Several species self-organize into large groups to leverage vital behaviors such as foraging, construction, or predator evasion. Ants, for instance, find shortest paths to food resources by depositing pheromones, bees indicate direction and distance to flower meadows through waggle dances in the hive, and fish display evasive maneuvers to escape predators. These three examples illustrate a collective problem-solving ability that leverages the cognition and actions of individually limited organisms.
With the advancement of robotics and automation, engineered multi-agent systems have been inspired to achieve similarly high degrees of scalable, robust, and adaptable autonomy through decentralized and dynamic coordination. Scientists have demonstrated ground-based collective transport, construction, and self-assembly, in some cases with several hundred robots. Multiple aerial swarms fly complex maneuvers, some of them even with little external assistance. Small robot teams, although with limited autonomy, have been engaged to assist in search and rescue missions at sea, sample oceanic data, and find unknown deep-sea species.
Overall however, robot swarms have been most successfully demonstrated in two-dimensional (2D) space or with partial assistance from central controllers and external tracking. In addition, many more demonstrations of self-organized collectives exist above-ground as opposed to the less explored underwater domain, which is particularly challenging because it often precludes traditional communication methods such as radio and GPS signals. Few underwater swarms exist and achieve limited coordination complexity and scale because they rely on explicit message passing.
In this dissertation, I introduce a novel underwater robot collective, the Blueswarm, which realizes full 3D spatiotemporal coordination without any external assistance. Each Bluebot is equipped with four independently controllable fins and two wide-angle lens cameras for 3D locomotion and perception. The vision system is complemented by three LEDs, which encode information about direction, distance and heading, and facilitate implicit coordination among robots. In the bioinspired design process, I pursued simplicity in both hardware and software to enable real-time onboard multi-robot tracking for local decision making followed by swift action. Blueswarm is the first 3D underwater collective that uses only local implicit vision-based coordination to self-organize.
Inspired by the dynamic and agile coordination of fish, I show that complex and dynamic 3D collective behaviors — synchrony, aggregation-dispersion, dynamic circle formation, search-capture, and escape — can be achieved by sensing minimal, noisy impressions of neighbors without any centralized intervention. To the best of my knowledge, this is the first significant demonstration of unsupervised and autonomous 3D collective coordination underwater.
Accompanied by a custom simulator, the Blueswarm platform gives researchers a much-needed tool to systematically develop and test algorithms for self-organzied 3D collective behaviors in the laboratory. The results of this dissertation provide insights into the power of implicit coordination and advance the potential for future underwater robots that display collective capabilities on par with fish schools for applications such as environmental monitoring and search in coral reefs and coastal environments. In addition, the Bluebots are also well suited as an experimental testbed for investigating natural collective behaviors and biomimicry, for example, studying the energy savings for different formations in schooling fish or the performance landscape of aquatic propulsion with a diverse set of caudal fins.
Citable link to this pagehttps://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37368416
- FAS Theses and Dissertations