The present study is an initial investigation of complex decision making performance. Specifically, this work investigated how individuals make decisions involving risk and uncertainty within a spatially focused route planning task involving multiple simulated unmanned vehicles and objectives. Forty-three participants were instructed to create twenty-four route plans, half of which included a combination of risky route suggestions and enhanced icons. Given the high working memory demands of developing a route for multiple vehicles, it was expected that enhanced visualizations, which provided redundant information on target priority, uncertainty, and deadline, would reduce demands on working memory and improve performance. Additionally, it was predicted that providing risky route suggestions would negatively impact performance, yet enhanced visualizations would reduce the effect of risky route suggestions. However, findings supported neither hypothesis; there was no performance difference in providing enhanced visualizations, and risky sub-optimal route suggestions actually improved the expected value of the submitted route. These results could have occurred due to multiple factors, including the multi-objective route planning task being too complex, or the scenarios lacking sufficient variability in expected value. An additional interpretation of these findings is that humans are really poor at performing these complex multi-objective tasks, and failed to comprehend the uncertainty and risk inherent in their decisions.
Efficient multi-agent task allocation for collaborative route planning with multiple unmanned vehicles
