AccessGrid Presentation - "Running with dinosaurs: fossils, physics and physiology"

Bill Sellers, Integrative Vertebrate Biology, Faculty of Life Sciences, University of Manchester, 2-3pm Room 1.10, Kilburn Building, The University of Manchester

An Access Grid presentation in conjunction with vizNET and the ACM SIGGRAPH University of Manchester Professional Chapter

Traditional techniques for reconstructing gait in fossil dinosaurs either involve complex animatronic machines or stop motion animation techniques. These techniques rely on a good knowledge of the skeletal anatomy and a familiarity with the range of locomotor styles seen in modern animals. These are mixed with a great deal of artistic skill and often produce visually stunning results. However using these approaches it is impossible to say whether the animal could actually have moved as portrayed. The movements used are anatomically possible but in all likelihood if the animal had actually tried to move like this then it would have fallen over. Even had it managed to stay upright it would not have minimised its cost of locomotion as living animals do. These difficulties can be overcome if we include both Newtonian physics and musculoskeletal physiology in conjunction with skeletal anatomy in our reconstructions. To do this we create a computer model of the musculoskeletal system of our target vertebrate fossil. The limbs and body are reconstructed as jointed segments, and the muscles and tendons are force generators that power the movement. This requires us to make estimates of various soft tissue parameters which are generally not preserved in the fossil record so we use a combination of phylogenetic and functional bracketing to estimate these values from living animals. The computer model is then imported into a physics simulator which solves the equations of motion so that the model moves appropriately given the forces applied by the muscles, by contact with the ground, and by gravity. Unfortunately such a model will not spontaneously walk or run so we use a genetic algorithm search procedure to find muscle activation patterns that optimise global parameters such as minimising energy cost or maximising speed. The end result is the generation of stable gait that is anatomically, physiologically and physically possible. At the same time the gait can represent an objective estimate of the most energetically efficient gait, or alternatively the fastest gait possible for a given animal. Sadly current technology does not produce gaits that look as good as the more artistic techniques and this new technique highlights the uncertainty inherent in all attempts at gait reconstruction. However ultimately it is a very powerful approach for the scientific understanding of dinosaur gait and we predict that as the technology advances it will also find a place assisting more artistic reconstructions. http://www.animalsimulation.org