1. This paper “presents an approach to improve the search efficiency for near-optimal motion synthesis using motion graphs.” (Abstract)
2. Many successful techniques for realistic motion synthesis are based on motion graphs, but the search complexity for even a near-optimal solution increases exponentially as the length of the desired sequence increases.
3. Apply bi-directional A* search on the motion graph, expanding two search trees simultaneously; one from the beginning and one from the end of the motion sequence.
4a. Motion is synthesized with significant speedup from existing methods, a maximum of 81x speedup from unidirectional search under favorable conditions.
4b. A sketch interface is also demonstrated where users can alter the trajectory of motion using intuitive, simple strokes instead of memorizing lists of mappings between stoke patterns and motions. Could provide an alternative to joysticks, mice, keyboards etc for input.
5. Website including video http://graphics.ucsd.edu/~wanyen/Wan-Yen_Lo/BiSearch.html