Decentralized Robotic Foraging Through Local Interactions

Dillan Wilson*, Nathaniel Pearson, John Little, and Yu Gu
Benjamin M. Statler College of Engineering and Mineral Resources

Presentation No.: 70

Assigned Category (Presentation Format): Engineering (Poster Presentations)

Student’s Major: Mechanical Engineering

In nature, a few examples of local foraging interactions between collaborating agents are chemical pheromones and wiggle dances with ants and bees respectively. This research strives to replicate simple natural interaction rules for robotic swarms that lead to efficient decentralized foraging behavior. In this work, agents locally observe other agents traveling home while transporting a resource and change their movement to an opposing direction to increase their chances of finding resources. Solving the robotic swarm foraging problem is important for applications such as search and rescue, environmental clean up, and planetary resource gathering. Swarms leverage collective intelligence to operate more efficiently than multiple non-collaborative robots and more flexibly than centrally organized robots by adapting to their environment. Swarm foraging is not reliant on a single robot and remains robust to multiple unit failures. A simulation has been developed where the behavior of the swarm can be influenced by altering key robot parameters: range of wiggle motion, conformity, and conformity duration. These adaptable parameters can create emergent behavior to increase foraging efficiency depending on the environment.

Funding: NSF, Award # 1851815

Program/mechanism supporting research/creative efforts: the WVU Robotics REU (Yu Gu & Jason Gross)