Army ants are unlike any other ant species. They are nomadic and therefore do not build permanent nests. Instead, between two episodes of sometimes daily migration, they build temporary nests called ?bivouacs? by attaching themselves to each other. In this way they build functional structures solely out of their interconnected bodies. Army ants can assemble and disassemble bivouacs of up to a million individuals in less than an hour and can do so in almost any condition in which they find themselves. These incredible feats of constructions are of great interest for engineers looking to create swarms of robots that could autonomously assemble themselves anywhere into any desired shape. Similar swarms could, for instance, self-assemble into habitats on Mars or in disaster areas without human intervention and regardless of the state of the local environment. This collaborative research brings together biologists and roboticists to achieve two goals: understand the principles of self-assembling construction in army ants, and adapt these principles to create a new generation of robots capable of self-assembling into any desired functional structure, even in unpredictable environments. The project will also give students from K12 to Ph.D. an opportunity to learn how biological structures build themselves out of smaller units, and how fundamental knowledge of natural processes can lead to new technological developments and applications in engineering.
The project has three complementary components. In Component 1, the researchers will perform field experiments to determine the rules used by army ants to self-assemble into functional structures. These studies will combine computer vision-assisted behavioral observations to measure the individual behaviors of the ants and high-definition 3D imaging using a custom-designed CT-scanner to characterize the organization and dynamics of the structure under construction. In Component 2, the result of the field experiments will be used to generate a multi-agent mathematical model and a physics-based simulation of the ant behaviors. The focus will be to design generalizable agent abstractions that allow for mathematical analysis to determine what forms of individual rules lead to correct and efficient collective outcomes as determined by the functional goal (e.g. formation of bivouac), and how modification of individual sensing and coordination capabilities affect the colony capability. Finally, in Component 3, the researchers will design a self-assembling robotic swarm with at least 30 robots. The swarm will be capable of building functional structures in unknown environments, through climbing and attaching to each other. These robots will use similar principles as army ants for collective control and allow complex 3D ?organic? self-assembled structures. The goal is not to mimic ant morphology, but instead demonstrate novel robot designs that use embodied intelligence and bio-inspired control, to achieve similarly adaptive structures while also allowing simplicity and large-scale manufacturability.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
