Autonomous area exploration by decentralized robot swarms is constrained in practice by finite energy budgetsthat determine operational lifetime. Existing algorithms — random walk, frontier-based exploration, andant-colony inspired methods — treat energy as an external constraint rather than a first-class decision parameter, resulting in suboptimal coverage trajectories and premature agent depletion. This paper introduces EADE (Energy-Aware Dynamic Exploration), a decentralized swarm algorithm that continuously adapts its frontierselection distance, communication frequency, and motion strategy to each robot's instantaneous energy fractionepsilon (t) = E (t) /E₀. EADE integrates five mechanisms: (1) energy-proportional frontier targeting via acontinuous scoring function, (2) compressed delta-sharing communication at O (1) bandwidth, (3) virtualrepulsion-based collision avoidance without dedicated safety messages, (4) Lévy flight bursts for escape fromlocally explored regions, and (5) spiral fallback for systematic coverage recovery. We evaluate EADE againstthree established baselines across a full-factorial 480-run experiment (4 algorithms x 4 robot counts x 2 obstacledensities x 15 independent replicates). EADE achieves mean T₉0 = 57. 8 steps (SD = 63. 8), compared to 89. 6 (frontier-based), 173. 9 (ant colony), and 210. 6 (random walk). Energy consumption is reduced by 53. 3%relative to random walk (Cohen's d = 1. 24). Zero inter-robot collisions are recorded across all conditions. Effectsizes are large for convergence speed (d = 1. 27 vs. random walk, d = 1. 06 vs. ant colony). Results arestatistically significant (p < 0. 001 for all comparisons on T₉0; p = 0. 026 for final coverage vs. frontier-based). The algorithm operates without a global map, requires no offline training, and introduces a formally definedenergy-to-ambition mapping not present in prior swarm exploration literature.
Parvesh (Sat,) studied this question.