Evolving Robots on Easy Mode: Towards a Variable Complexity Controller for Quadrupeds

The complexity of a legged robot's environment or task can inform how specialised its gait must be to ensure success. Evolving specialised robotic gaits demands many evaluations - acceptable for computer simulations, but not for physical robots. For some tasks, a more general gait, with lower optimization costs, could be satisfactory. In this paper, we introduce a new type of gait controller where complexity can be set by a single parameter, using a dynamic genotype-phenotype mapping. Low controller complexity leads to conservative gaits, while higher complexity allows more sophistication and high performance for demanding tasks, at the cost of optimization effort. We investigate the new controller on a virtual robot in simulations and do preliminary testing on a real-world robot. We show that having variable complexity allows us to adapt to different optimization budgets. With a high evaluation budget in simulation, a complex controller performs best. Moreover, real-world evolution with a limited evaluation budget indicates that a lower gait complexity is preferable for a relatively simple environment.

[1]  J. Estremera,et al.  Quadrupedal Locomotion: An Introduction to the Control of Four-legged Robots , 2006 .

[2]  Joshua Evan Auerbach,et al.  Environmental Influence on the Evolution of Morphological Complexity in Machines , 2014, PLoS Comput. Biol..

[3]  Denis Fisseler,et al.  Learning in a High Dimensional Space: Fast Omnidirectional Quadrupedal Locomotion , 2006, RoboCup.

[4]  Kyrre Glette,et al.  Overcoming Initial Convergence in Multi-objective Evolution of Robot Control and Morphology Using a Two-Phase Approach , 2017, EvoApplications.

[5]  Julian Togelius,et al.  Evolution of a subsumption architecture neurocontroller , 2004, J. Intell. Fuzzy Syst..

[6]  I. Harvey,et al.  Do Not Disturb : Recommendations for Incremental Evolution , 2010 .

[7]  Joshua D. Knowles,et al.  Bounded archiving using the lebesgue measure , 2003, The 2003 Congress on Evolutionary Computation, 2003. CEC '03..

[8]  Erik D. Goodman,et al.  Genetic Programming-Based Automatic Gait Generation in Joint Space for a Quadruped Robot , 2010, Adv. Robotics.

[9]  Philip K. McKinley,et al.  A Comparison of Multiobjective Algorithms in Evolving Quadrupedal Gaits , 2016, SAB.

[10]  Antoine Cully,et al.  Robots that can adapt like animals , 2014, Nature.

[11]  Kenneth O. Stanley,et al.  Confronting the Challenge of Quality Diversity , 2015, GECCO.

[12]  Jie Yan,et al.  A Review of Gait Optimization Based on Evolutionary Computation , 2010, Appl. Comput. Intell. Soft Comput..

[13]  Huosheng Hu,et al.  GA-BASED GAIT GENERATION OF SONY QUADRUPED ROBOTS , 2003 .

[14]  Kyrre Glette,et al.  Multi-objective evolution of fast and stable gaits on a physical quadruped robotic platform , 2016, 2016 IEEE Symposium Series on Computational Intelligence (SSCI).

[15]  Hod Lipson,et al.  Evolving robot gaits in hardware: the HyperNEAT generative encoding vs. parameter optimization , 2011, ECAL.

[16]  Auke Jan Ijspeert,et al.  Central pattern generators for locomotion control in animals and robots: A review , 2008, Neural Networks.

[17]  Stéphane Doncieux,et al.  Overcoming the bootstrap problem in evolutionary robotics using behavioral diversity , 2009, 2009 IEEE Congress on Evolutionary Computation.

[18]  Kyrre Glette,et al.  Self-Modifying Morphology Experiments with DyRET: Dynamic Robot for Embodied Testing , 2019, 2019 International Conference on Robotics and Automation (ICRA).

[19]  Kyrre Glette,et al.  Real-world evolution adapts robot morphology and control to hardware limitations , 2018, GECCO.