Bio-inspired Robot Swarms

Swarmbots is a project on development of algorithms for automation of a swarm of autonomous mobile robots. The robots perform various tasks through ‘collective decision making’. Multi-robot systems become increasingly complex with an increase in the number of robots. In this project, a swarm of robots is entirely decentralized and each robot knows only the approximate relative position of its neighbours. Furthermore, the bots do not rely on inter-robot communication, localization or memory, and possess only minimal local sensing & processing capabilities. They perform tasks collectively through ‘swarm intelligence’ algorithms.

Swarmbots: All tasks are decentralized, i.e. each robot takes its own decisions. The images show the emergent collective behaviours of Area coverage/ exploration (Left), Contamination removal (Centre) and Foraging (Right)

The project attempts to break down complex coordinated tasks into simpler sub-tasks such as Agregation, Line-formation, Shape-formation, Area-Coverage, Flocking, Following, etc. The project was started in 2019 with RasPi-controlled differential drive robots, and was moved to simulations in 2020.

The project implemented the following tasks:

1. Area Search: The robots disperse and collectively explore an area.
2. Foraging/ Resource Gathering: The robots search for & collectively transport resources to a ‘nest’ at an unknown location.
3. Contamination Removal: The robots create a perimeter around gradually growing contaminations and neutralize them

These tasks were implemented weighted combinations of basic swarm-behaviours such as aggregation, dispersion, exploration, line formation, shape formation, surrounding items and consensus.

Role in Project: Project Lead. Developed the Algorithms, Simulation

[GitHub Repo]

[Talk YouTube link]

[Nature Scientific Reports submission]