IIIT-Bangalore · Project Intern
Swarm Intelligence for Multi-Robot Blanket Coverage
A novel swarm algorithm achieving 92% spatial coverage and 40% fewer iterations than Q-learning baselines.
Period
Jan – May 2023
Tags
Swarm · Multi-Agent · Benchmarking
Highlights
- 0192% spatial coverage efficiency for multi-robot blanket coverage
- 0240% fewer convergence iterations vs. Q-learning baseline
- 03Maintained 95% of Q-learning performance with significantly lower compute
Write-up
Benchmarked against the Virtual Force Approach and Q-Learning, the proposed algorithm prioritized fast convergence with bounded performance loss — a tradeoff well-suited to resource-constrained swarm deployments.
Links