Abstract: To implement the motion control of virtual humans in complicated environment,an approach to optimizing virtual human postures for avoiding obstacles is presented,which is based on the artificial potential field.Firstly,the artificial potential field is created according to the target and obstacle positions in the virtual environment,and then a collision-free path is planned for the end-effector of virtual human motion chain.Secondly,the computational problem of inverse kinematics for each posture of the end-effector along the path is described and resolved by the comfort level function optimization problem with posture and collision-free constraints.Finally,a series of collision-free continuous postures of human movement can be computed,and then the obstacle avoidance motion control is achieved.An example is given to validate the approach.The approach can be applied to virtual human animation and ergonomics simulation in the complicated environment.