Learning optimal gait parameters and impedance profiles for legged locomotion

The successful execution of complex modern robotic tasks often relies on the correct tuning of a large number of parameters. In this paper we present a methodology for improving the performance of a trotting gait by learning the gait parameters, impedance profile and the gains of the control architecture. We show results on a set of terrains, for various speeds using a realistic simulation of a hydraulically actuated system. Our method achieves a reduction in the gait's mechanical energy consumption during locomotion of up to 26%. The simulation results are validated in experimental trials on the hardware system.

[1]  Bram Vanderborght,et al.  Development of a compliance controller to reduce energy consumption for bipedal robots , 2008, Auton. Robots.

[2]  Darwin G. Caldwell,et al.  LOCAL REFLEX GENERATION FOR OBSTACLE NEGOTIATION IN QUADRUPEDAL LOCOMOTION , 2013 .

[3]  Jan Peters,et al.  Policy Search for Motor Primitives in Robotics , 2008, NIPS 2008.

[4]  S. R. Jammalamadaka,et al.  Directional Statistics, I , 2011 .

[5]  Martijn Wisse,et al.  A Disturbance Rejection Measure for Limit Cycle Walkers: The Gait Sensitivity Norm , 2007, IEEE Transactions on Robotics.

[6]  R. McNeill Alexander,et al.  Principles of Animal Locomotion , 2002 .

[7]  Stefan Schaal,et al.  Learning Policy Improvements with Path Integrals , 2010, AISTATS.

[8]  Stefan Schaal,et al.  Learning variable impedance control , 2011, Int. J. Robotics Res..

[9]  Jan Peters,et al.  Reinforcement learning in robotics: A survey , 2013, Int. J. Robotics Res..

[10]  Andrew Howard,et al.  Design and use paradigms for Gazebo, an open-source multi-robot simulator , 2004, 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE Cat. No.04CH37566).

[11]  Olivier Sigaud,et al.  Path Integral Policy Improvement with Covariance Matrix Adaptation , 2012, ICML.

[12]  Dorothea Heiss-Czedik,et al.  An Introduction to Genetic Algorithms. , 1997, Artificial Life.

[13]  Peter Stone,et al.  Machine Learning for Fast Quadrupedal Locomotion , 2004, AAAI.

[14]  Navinda Kottege,et al.  Autonomous navigation of hexapod robots with vision-based controller adaptation , 2017, ICRA 2017.

[15]  Manuel G. Catalano,et al.  Variable impedance actuators: A review , 2013, Robotics Auton. Syst..

[16]  Evangelos A. Theodorou,et al.  An iterative path integral stochastic optimal control approach for learning robotic tasks , 2011 .

[17]  Ferdinando Cannella,et al.  Design of HyQ – a hydraulically and electrically actuated quadruped robot , 2011 .

[18]  J. Buchli,et al.  Learning Compliant Locomotion on a Quadruped Robot , 2014 .

[19]  Jessica K. Hodgins,et al.  Adjusting step length for rough terrain locomotion , 1991, IEEE Trans. Robotics Autom..

[20]  Martijn Wisse,et al.  Running with improved disturbance rejection by using non-linear leg springs , 2011, Int. J. Robotics Res..

[21]  William H. Press,et al.  Numerical Recipes in Pascal , 2007 .

[22]  Darwin G. Caldwell,et al.  Probabilistic Contact Estimation and Impact Detection for State Estimation of Quadruped Robots , 2017, IEEE Robotics and Automation Letters.

[23]  Stefan Schaal,et al.  Reinforcement learning of motor skills in high dimensions: A path integral approach , 2010, 2010 IEEE International Conference on Robotics and Automation.

[24]  Yasemin Altun,et al.  Relative Entropy Policy Search , 2010 .

[25]  Darwin G. Caldwell,et al.  A reactive controller framework for quadrupedal locomotion on challenging terrain , 2013, 2013 IEEE International Conference on Robotics and Automation.

[26]  Alin Albu-Schäffer,et al.  The KUKA-DLR Lightweight Robot arm - a new reference platform for robotics research and manufacturing , 2010, ISR/ROBOTIK.

[27]  Yasuhiro Fukuoka,et al.  Adaptive Dynamic Walking of a Quadruped Robot on Irregular Terrain Based on Biological Concepts , 2003, Int. J. Robotics Res..

[28]  Melanie Mitchell,et al.  An introduction to genetic algorithms , 1996 .

[29]  Yishay Mansour,et al.  Policy Gradient Methods for Reinforcement Learning with Function Approximation , 1999, NIPS.