Automatic derivation of curved human walking trajectories from synthetic vision

The scope of this paper is to propose a system for the automatic derivation of a human curved walking trajectory from the analysis provided by its synthetic vision module. The study context is a planar environment with still and sparse foothold locations. A general methodology associates the two low-level modules of vision and walking with a planification module which establishes the middle term path from the knowledge of the visualized environment. The planification is made under the constraint of minimizing the distance, the speed variation and the curvature cost. Moreover, the planification may trigger the alternate walking motion whenever the decreasing in curvature cost is higher than the associated increasing in speed variation cost due to the corresponding halt and restart. Finally, the characteristic of the next step location is derived in the context of a constant velocity walking motion along a circular trajectory