learning dynamic manipulation skills under
play

Learning Dynamic Manipulation Skills under Unknown Dynamics with - PowerPoint PPT Presentation

Learning Dynamic Manipulation Skills under Unknown Dynamics with Guided Policy Search Sergey Levine Pieter Abbeel UC Berkeley UC Berkeley Team TROOPER: Lockheed Martin, University of Pennsylvania, Philipp Krahenbuhl, Stanford University


  1. Learning Dynamic Manipulation Skills under Unknown Dynamics with Guided Policy Search Sergey Levine Pieter Abbeel UC Berkeley UC Berkeley

  2. Team TROOPER: Lockheed Martin, University of Pennsylvania, Philipp Krahenbuhl, Stanford University Rensselaer Polytechnic Institute

  3. general-purpose neural network controller +

  4. policy search (RL) complex dynamics complex policy HARD supervised learning complex dynamics complex policy EASY trajectory optimization complex dynamics complex policy EASY trajectory optimization supervised learning

  5. Trajectory Optimization prob guided policy search

  6. Trajectory Optimization • locally linear dynamics approximate solution using iterative LQR (similar to extended Kalman filter) • locally quadratic cost • Gaussian distribution

  7. Trajectory Optimization

  8. Trajectory Optimization new old

  9. Trajectory Optimization

  10. Trajectory Optimization

  11. Trajectory Optimization

  12. Guided Policy Search see Levine & Koltun, ICML 2014

  13. Concluding Comments • simple linear dynamics model • fast, simple, standard LQR solver • can handle contacts despite linear model • fit very complex policies with guided policy search

Recommend


More recommend