A Reproducible Framework for Evaluating Autonomic Swarm Recruitment: Enhancing MegaSwarm with Systematic Metrics, Visualization, and Performance Analysis

Swarm robotics provides scalability in task allocation due to decentralized control but the performance drops when robots act in the absence of organized coordination, especially when the environment is clustered or unbalanced. The biologically inspired behaviors in SwarmSim2 and MegaSwarm simulators were pheromone trails, recruitment, autonomic pulse communications and fault-tolerant sensing. Although these studies were able to define significant concepts, they were mostly restricted to case-based illustrations and statistical observation of one-off situations. In this paper, a research quality experimental framework is built on top of MegaSwarm that allows systematic and reproducible evaluation of recruitment strategies. Automation of batch sweeps of recruitment parameters, standardized metrics logging, and reproducible visualization output in the form of heatmaps, pheromone fields and time series charts have been added. The four recruitment modes, Off, OneResponder, MultiResponders and Blackboard, are tested with different swarm sizes and spatial layouts. Baseline experiments performed partial completion of the tasks (12-18 of 20 sites, mean scores plateaued at 0.6-0.8), whereas recruitment strategies increased coverage and completion times dropped significantly with Blackboard showing near total completion. This framework will turn MegaSwarm into an analytical tool to quantitatively study trade-offs between time, energy and communication in swarm recruitment.