Extreme Scalability: Designing Interfaces and Algorithms for Soldier-Robotic Swarm Interaction

Abstract

Abstract : In theory, autonomous robotic swarms can be used for critical Army tasks (i.e., accompanying convoys); however, the Soldier controlling the swarm must be able to monitor swarm status and correct actions, especially in disrupted or degraded conditions. For this two-year Director's Research Initiative (DRI), we designed metacognition algorithms and Soldier-swarm display concepts to allow Soldiers to efficiently interact with a robotic swarm participating in a representative convoy mission. We used a potential field approach for swarm control because it scales easily to large heterogeneous swarms and allows users to dynamically alter swarm behavior by adjusting field parameters. The Soldier-swarm interface displayed swarm and convoy geospatial position; swarm health and communication; and convoy status information, using visual, auditory, and tactile combinations. We measured swarm metacognition by determining the proportion of time the simulated swarm could maintain a specific orbital ring around the convoy over six terrains in 13-min scenarios. We tested interface effectiveness in a laboratory study using 16 male Marines (volunteers) with a mean age of 19 years. The metacognition results showed that the swarm could maintain the pre-defined dispersion more than 85% of the time in each terrain. Using multimodal displays, Soldier workload decreased and performance increased (i.e., response time reduced).