Identification of Stochastic Hybrid System Models

1. Identification of StochasticHybrid System Models MCE-4.7 S. Shankar Sastry 5 yrs 5 sastry@eecs.berkeley.edu Abstract: Much work remains to develop an empirically-validated theory enabling systematic design and control of multi-leg terrestrial robots. The control task is daunting, as a large number of degrees-of-freedom must be rapidly and precisely coordinated in the face of state and environmental uncertainty. In spite of these difficulties, animals at all levels of complexity have mastered the art of rapid legged locomotion over complex terrain at speeds far exceeding those of comparable robotic platforms, for instance those under development in MAST. Our results in previous years provide a framework in which reduced-order hybrid dynamical models can be used to study rhythmic legged locomotion. In ongoing work we seek to translate these theoretical advancements into practical tools for data-driven modeling of aperiodic behaviors, automated synthesis of gaits and maneuvers, and design of robust multi-leg gaits. Journal Papers (submitted, in review): [MCE-13-4.7-5] S. A. Burden, H. Gonzalez, R. Vasudevan, R. Bajcsy, and S. S. Sastry. Metrization and Simulation of Controlled Hybrid Systems. In review, IEEE Transactions on Automatic Control, 2013. [MCE-13-4.7-6] S. A. Burden, S. Revzen, and S. S. Sastry. Model Reduction Near Periodic Orbits of Hybrid Dynamical Systems. In review, IEEE Transactions on Automatic Control, 2013. Conference Proceedings: [MCE-11-4.7-1] S. A. Burden, H. Gonzalez, R. Vasudevan, R. Bajcsy, and S. S. Sastry. Numerical Integration of Hybrid Dynamical Systems via Domain Relaxation. IEEE Conference on Decision and Control, 2011. [MCE-11-4.7-2] S. A. Burden, S. Revzen, and S. S. Sastry. Dimension reduction near periodic orbits of hybrid systems. IEEE Conference on Decision and Control, 2011. [MCE-12-4.7-1] S. A. Burden, H. Ohlsson, and S. S. Sastry. Parameter identification near periodic orbits of hybrid dynamical systems. IFAC Symposium on System Identification, 2012. [MCE-13-4.7-1]S. A. Burden, S. Revzen, T. Y. Moore, S. S. Sastry, and R. J. Full. Using reduced-order models to study dynamic legged locomotion: Parameter identification and model validation. Annual Meeting of the Society for Integrative and Comparative Biology, 2013. [MCE-13-4.7-2] S. A. Burden and S. S. Sastry. Reduction and Identification for Hybrid Dynamical Models of Terrestrial Locomotion. Proceedings of the SPIE Conference on Defense, Security, and Sensing, 2013. [MCE-13-4.7-3] S. A. Burden, S. Revzen, and S. S. Sastry. From Anchors to Templates: Exact and Approximate Reduction in Models of Legged Locomotion. Dynamic Walking, 2013. [MCE-13-4.7-4] S. Revzen, S. A. Burden, and S. S. Sastry. Pinned Equilibria Provide Robustly Stable Multilegged Locomotion. Dynamic Walking, 2013. [MCE-14-4.7-1]S. A. Burden, S. S. Sastry, and R. J. Full. Optimization for models of legged locomotion. Annual Meeting of the Society for Integrative and Comparative Biology, 2014 Collaborations/Transitions/Leveraged Efforts: Active collaborations with Prof. Robert Full (UCB, MCE) and Prof. Daniel Goldman (GATech, MCE). Past collaborations with Prof. Ronald Fearing (UCB, MCE).