How Many Robots? Group Size and Efficiency in Collective Search Tasks

Abstract

Abstract. This paper presents a quantitative analysis of the tradeoffs between group size and efficiency in collective search tasks that considers both the timesensitive nature of search completion and the system operating cost. First, the search task is defined and a performance metric is presented that can account for all of the costs associated with the task. Next, for both random and coordinated search strategies, analytical expressions are derived that can be used to predict optimal system performance bounds given a particular task description, and the performance benefit of using coordinated search is shown to be dependent on the relative values of the different cost components. Finally, an embodied computer simulation is used to support the analytical results, suggesting that the assumptions involved in their derivation are sound.

Energy-Time Efficiency in Aerial Swarm Deployment

“…The tradeoff between group size and efficiency was examined by Hayes [6] in a search task. A multi-objective performance function was used incorporating search time, energy and robot initialisation costs.…”

“…Alternatively, multi-objective functions are used, combining multiple weighted metrics, e.g. Hayes [6]. However, this is not straightforward due to the choice of metrics, weighting and formulation.…”

“…Rapid deployment is desired to expedite tasks such as disaster mitigation. However, time and energy are not independent, and often there is a tradeoff [13,6,10]. Previous research considered only ground robots.…”

Distributed Adaptation in Multi-robot Search Using Particle Swarm Optimization

“…This work has been relatively disjoint thus far, with most studies focusing on a particular scenario which is not explicitly connected to other related work. The cost of using additional robots in a search task was explored and tested with simulation [7]. Detailed analysis has been done for swarms following a gradient [16].…”

“…In 2001, a contest at the International Joint Conference on Artificial Intelligence on collective robotic urban search and rescue [17] prompted some research on the topic [10]. Other publications explore multi-robot search strategies in simulation [6], for infrared tracking with simulation and real robots [7], and for odor source localization with real robots [8].…”