Adaptive Walk-Kick on a Bipedal Robot

Abstract

Using the NAO robot as a testbed, we propose a walk-kick framework that can generate a kick trajectory with an arbitrary direction without prior input or knowledge of the parameters of the kick in the midst of walking while still guaranteeing reaching a reference trajectory. The walk-kick uses kick interpolators from a dynamic kick engine and the walk trajectories generated from adaptive walking engine to generate motions in any direction that allow a robot to reach its destination while also allowing it to move the ball in further distances without transitioning in different states to accommodate both tasks. The system has been extensively tested on the physical robot, taking into account ten different target angles. The stability and reliability of each kick has been evaluated 30 times for each kick motion trajectory while performing demanding motions. Results show that our proposed walk-kick framework and its integration is reliable in terms of the kick directions and stability of the robot overall (<1% falling rate), and our experiments verify that the walk-kick trajectories were consistent with an average absolute bearing of <6$$^\circ $$ within any given direction.