Optimal viewpoint selection for indoor drones using PSO and Gaussian Processes
Taisei Yokomatsu, Kosuke Sekiyama · Society of Instrument and Control Engineers of Japan · 2021
In this study, we developed an indoor drone photographer which takes a well-composed photo of a group of people from the air. By combining the Gaussian process and the PSO(Particle Swarm Optimization) search in the three-dimensional space search method, it becomes a wide-area search that does not fall into the local world, and it is possible to efficiently predict the heuristic optimum viewpoint in a short time. In addition, the ideal arrangement can be freely changed by using the method in our previous research to evaluate the composition balance of photographs.