Human perception and prediction of robot swarm motion
Abstract
Great strides have been made in designing autonomous swarms of robotic agents, but there is a performance gap in complex situations and rapidly changing circumstances that human-agent teaming can fill. To design swarm control algorithms that are intuitive for human teammates to learn and use, it is necessary to understand how humans perceive groups of objects and predict group motion. To that end, we presented participants with groups of triangles as schematic representations of robot swarms and asked participants to predict where the center of the moving swarm would be in the future. We previously found that judgements of static swarm centers were biased by the direction of the apparent swarm heading, being attracted to the heading point when it was inside the group and repelled by it when it was outside of the group. Here we show that predictions of the center of moving swarms' positions were influenced by both the type of motion of the swarm and the control algorithm between individual robot members, with more accurate temporal and spatial responses when swarm movement was sinusoidal rather than parabolic and when the all robot members moved as one unit, rather than following a leader robot. These findings have implications for the design of swarm control systems, giving guidance on what biases to account for when designing the most intuitive controls. Also, there are theoretical implications for ensemble perception in psychology, as it runs counter to the prevailing theory that mean position judgements and mean motion judgements are mentally computed easily and without bias.
- Publication:
-
Micro- and Nanotechnology Sensors, Systems, and Applications XI
- Pub Date:
- May 2019
- DOI:
- 10.1117/12.2517776
- Bibcode:
- 2019SPIE10982E..26C