This example shows a 3D animated dog trot which was extracted video. The 2D Markers trajectories are extracted from a rectified video of a nodal pan camera. The 3D motion is reconstructed by guiding a physically based simulation with 2D trajectories. While the motion is largely planar, the degrees of freedom of the dog model combine with dynamic effects (e.g., inertia of different body parts) to produce subtle out of plane motions, such as spinal flexion. This work has been done with Paul Kry (now at McGill University). The original video footage has been collected by the National Museum of Natural History in Paris, France.
Locomotion is an energetically expensive activity, and it is
possible that animals use their bodies in efficient ways to produce
that dependent on speed and task (for instance, a
slow gate to search for shelter or a fast gate to escape a
predator). The idea behind this work is that locomotion strongly
depends on the morphology of the animal, and we can deduce a lot of
information about plausible modes of locomotion just by observing the
natural vibrations of an animal's musculoskeletal structure.
This video shows a user defined controller which produces locomotion in a dynamic environment. This work has been done with Gregory Clauzel, who is now at EdenGames. The virtual animal is able to walk over and interact with dynamic obstacles, while maintaining balance using additional forcing inspired from a ZMP approach.