Views planning is the task of computing the sensor poses so that a target entity (scene or object) will be completely observed. In this project, our research group, named as «the active vision research group» is studying the problem as well as developing algorithms to adders it. Our last contributions include the proposition of deep learning-based method for view planning, where we have found, that these new algorithms are capable of achieve similar result that previous algorithms in shorter computation times.
Last experiments have been run using blensor simulator, even that such simulator does not provide real-time experimentation, it is easy to intall and avoids the use of additional software.
- Yervilla-Herrera, H., Vasquez-Gomez, J.I., Murrieta-Cid, R. et al. “Optimal motion planning and stopping test for 3-D object reconstruction“, Intel Serv Robotics (2018). https://doi.org/10.1007/s1137
- Vasquez-Gomez, J. Irving, et al. “Tree-based search of the next best view/state for three-dimensional object reconstruction.” International Journal of Advanced Robotic Systems 15.1 (2018).
(Open Access) https://doi.org/10.1177/1729881418754575