Human Locomotion

Our first simulation attempts were looking at human bipedalism. This has several advantages. Firstly human bipedalism is extremely well studies experimentally. This means that there is plenty of data around for checking our simulations. Secondly human experiments are very easy to do. Human volunteers are generally very obliging and do exactly as you ask. Anyone who has ever worked with animals will appreciate that this is not the case with non-humans. Thirdly there are a lot of theoretical studies of human locomotion which makes model building much easier. Our interest in this case was the energetic efficiency of bipedalism since this is often quoted as the main advantage that our early human ancestors had over the other great apes.

This simulatant is based on Dynamechs and uses a finite-state central pattern generator. It has been written up in a couple of publications (Sellers et al. 2003, 2004) which can be requested on the publications page. The key results were that the simulator was stable and produced accurate estimates of the metabolic energy cost of locomotion at low speeds. The results for running and skipping have not yet been written up but although they are stable, their current energy costs are very high - probably due to the lack of elastic storage in the model. The 'ankle-walking' animation is shown to illustrate that the process is able to generate (and evaluate) novel gaits.

Here are some example gait animations:

References

• Sellers WI, Dennis LA, Wang W, Crompton RH. Evaluating alternative gait strategies using evolutionary robotics. Journal of Anatomy. 2004 204: 343-351.
• Sellers WI, Dennis LA, Crompton RH. Predicting the metabolic energy costs of bipedalism using evolutionary robotics. Journal of Experimental Biology. 2003 206:1127-36.