Exploring Grammatical Evolution for Horse Gait Optimisation

Physics-based animal animations require data for realistic motion. This data is expensive to acquire through motion capture and inaccurate when estimated by an artist. Grammatical Evolution (GE) can be used to optimise pre-existing motion data or generate novel motions. Optimised motion data produces sustained locomotion in a physics-based model. To explore the use of GE for gait optimisation, the motion data of a walking horse, from a veterinary publication, is optimised for a physicsbased horse model. The results of several grammars are presented and discussed. GE was found to be successful for optimising motion data using a grammar based on the concatenation of sinusoidal functions.