Section: New Results

Optimization of transition paths

Participants : Leonard Jaillet, Stephane Redon.

Motion planning methods allow producing initial paths which represent transitions from one given conformation to another. However, these paths are typically suboptimal because of the probabilistic nature of the search strategy. Hence, it is necessary to develop tools to locally increase the path quality of the solution generated during the first phase. We have proposed several methods to address such a problem. One method developed is a variant of a state-of-the-art approach called nudged elastic band (NEB). It optimizes a set of intermediate images along the path, such that each image finds the lowest energy possible while maintaining equal spacing to neighboring images. Another technique we proposed is to rely on an equivalent of the shortcutting technique developed in Robotics motion planning, but applied to the context of energy landscape. Finally, we also have complemented these methods with additional tools to do simple path edition such as cutting or thinning paths.