Balanced Walking with Capture Steps Marcell Missura and Sven Behnke Autonomous Intelligent Systems, Computer Science, Univ. of Bonn, Germany {missura,behnke}@cs.uni-bonn.de http://ais.uni-bonn.de Abstract. Bipedal walking is one of the most essential skills required to play soccer with humanoid robots. Superior walking speed and stability often gives teams the winning edge when their robots are the first at the ball, maintain ball control, and drive the ball towards the opponent goal with sure feet. In this contribution, we present an implementation of our Capture Step Framework on a real soccer robot, and show robust omnidirectional walking. The robot not only manages to locomote on an even surface, but can also cope with various disturbances, such as pushes, collisions, and stepping on the feet of an opponent. The actuation is compliant and the robot walks with stretched knees. 1 Introduction For the RoboCup initiative, which has the goal of defeating the human world champions in the game of soccer by the year of 2050, it is of particular interest to conceive a bipedal walk with human-like capabilities. However, the complexity of the walking motion, the formulation of sufficiently simple models that account for balance, and the difficulties that arise from controlling a humanoid body with a high number of degrees of freedom within a feedback loop, make this task particularly difficult. While a number of sophisticated approaches exist that promise some degree of robustness, the RoboCup experience shows that state of the art algorithms do not find their way into the dynamic world of low-cost robots competing on the soccer field. One of the reasons for this is simply that the required sensors, high-precision actuators, and computational power are not available on custom built prototypes and affordable standard platforms that are used in robotic soccer games. The amount of expertise required to successfully integrate a complex algorithm into already complex soccer software in a real robot environment is also not a negligible factor. The Capture Step Framework [1] has been designed with the aforementioned limitations in mind. Using only postural information provided by motor encoders and an inertial measurement unit (IMU), it can produce a stable omnidirectional walk with push-recovery capabilities. Zero moment point control, foot placement, and step timing strategies are utilized simultaneously. The balance computations This work is supported by Deutsche Forschungsgemeinschaft (German Research Foundation, DFG) under grants BE 2556/6 and BE 2556/10. In Proceedings of 18th RoboCup International Symposium, Joao Pessoa, Brasil, July 2014. Best Science Paper Award.