ecoRacer: Can Human Find Optimal Solutions Faster for An NP-Complete Problem?
Collaborators: Ren, Y., Bayrak, A., and Papalambros, P.Y.Abstract
We developed this game to analyze how human players solve the optimal design and control problem embedded in a racer game: The player needs to finish the track within 36 seconds with as little energy consumption as possible. We then compare human results with the Efficient Global Optimization (Bayesian Optimization) algorithm. Results showed that some human beings are efficient at finding good solutions, yet most of them cannot get close to the true optimal solution.
Images
Game performance by human players (left) and the Efficient Global Optimization algorithm (right)
Related Papers
1. , Link