A criterion for assessing obstacle-induced environmental complexity in multi-robot coverage exploration.
Journal:
PloS one
PMID:
40378117
Abstract
In many applications, such as coverage exploration and search and rescue missions, accurately assessing environmental complexity is valuable for performance evaluation and algorithm adjustments. Despite this, in the context of multi-robot systems, quantifying environmental complexity caused by obstacles when using autonomous ground robots presents significant challenges. This research proposes a criterion for measuring environments' obstacle-induced complexity in the context of autonomous multi-robot coverage exploration. The criterion rates the environment's complexity numerically, where 0 denotes obstacle-free setups, and the value increases with obstacle-related effects, reaching a maximum of 1, representing the highest measurable complexity for the criterion. The proposed criterion is independent of robot hardware specifications and algorithm-specific aspects. Furthermore, it is independent of the environment's size and the ratio of the area occupied by obstacles, enabling comparisons across various environments. Statistical analysis shows the metric performs well both on average and in single-case comparisons.