Abstract
Robotic collectives (i.e., colonies and swarms) are applicable to a wide range of applications, including environmental monitoring, search and rescue, as well as infrastructure monitoring. The presented evaluation focuses on how two visualization designs impact human-collective team performance during a best-of-n sequential decision making task with colonies of 200 agents. Traditional visualizations present all the individual robots that encompass the entirety of the collective, which may cause the human operator to suffer from information overload which hinders understanding the collective’s current state, the reasoning behind actions, and associated predictive future outcomes. Interface designs that abstract the individual collective member details and present the collective’s state are needed to alleviate high workload and mitigate human error. The evaluation determined that an abstract visualization of the collective’s state produced better overall performance than the visualization that showed all the individual agents.