A novel locomotion synthesis and optimisation framework for insects
Journal: Computers and Graphics (Pergamon)
The motion modelling of insects is challenging because of their small size, high frequency movements and delicate dynamics model. This paper presents a novel framework which is specially tailored to synthesise this particular type of locomotion, inspired by the biological observation of the existence of Central Pattern Generator (CPG) which regulates the cyclic motion of creatures. In our method, the CPG is organically integrated with the modelling of dynamics with assistance of the abstract actuation mechanism, Virtual Model Control, to provide physically plausible results. To achieve realistic and natural motion, it is often not practical to manually set up the control parameters and we develop an automatic strategy to optimise such non-linear non-convex problem with the Covariance Matrix Adaptation which selects suitable candidates through an evolutionary process. This framework offers high level of control of characters' motion states, speed and directions, and is flexible to adapt characters' motion to perturbations and complex environments. © 2013 Elsevier Ltd. All rights reserved.