The Visual Categorization of Humanoid Movement as Natural

Abstract

We examined naturalness judgements of movements generated by different control strategies with the goal of producing natural looking movements on humanoid robots and virtual humans. We displayed motion clips on both a humanoid robot and a computer graphic character. Fourteen synthetic motion generation algorithms were developed based onhuman motor production theories. Human movement was also motion captured. Experimental results relate the actuator bandwidth, the level of computational complexity in motion generation, and the perceived naturalness of motion. In Experiment 1 it was found that for the humanoid robot, low ratings of naturalness were obtained for rapid movement. However Experiment 2 indicated that this effect appeared to not be due to specific features of the motor production techniques but instead due to a movement artifact generated by the humanoid robot.