At the 2009 Bristol SVP I presented work that derives from the dissertation research of Dr. Eric D. Wills. How much might one infer about how dinosaurs walked purely based on the kinematics of their limbs? Imagine a huge "configuration space" of possible limb poses. During the support phase of the walk cycle, each limb would trace a path through this configuration space. There are arbitrarily many paths, i.e. gaits, all of which are consistent with forward progress, hence the problem needs to be constrained. Straightforwardly searching for some "most efficient" (smoothest path of forward movement) is computationally impractical, but the search becomes more tractable if that configuration space is first reduced by some strong constraint. One such constraint derives from the fact that quadrupeds and bipeds have periods during their walk cycle where more than one leg is bearing the weight of the animal, hence more than one kinematic chain (series of joints from body to ground) are operating cooperatively during those periods of the gait. This greatly reduces the ambiguity, we found.

Dinosaurs, with their characteristic arrangement of joints from pectoral girdle to manus, and from pelvic girdle to pes, are not capable of strictly parasagittal limb movements, and hence have surprisingly restricted spaces of possible movements. We were able to find, however, efficient walking gaits for the semi-sprawling Triceratops, the more columnar hence parasagittal Apatosaurus, and the bipedal walk of Tyrannosaurus rex. We don't "can" the solution, and to demonstrate this, we even created a very non-biological biped (with knees and hips that had translational as well as rotational joints, and weird ankles as well).

A QuickTime Interactive based on the SVP presentation slides is provided here. Use the arrow keys or mouse clicks to advance slides (each slide with animation must terminate prior to advancing to the next slide).