Organization Welcome to • M/W 1:30 – 2:20 • Lectures, discussions (EEB 045) CSE 571 • Homework, project Probabilistic Robotics • Readings: • Papers Instructor: Dieter Fox • Chapters from Probabilistic Robotics Teaching Assistant: • Web page: Arun Byravan • http://www.cs.washington.edu/571 10/1/15 Probabilistic Robotics 2 SA-1 High-level View on Robot Goal of this course Systems • Provide an overview of problems / Sensor data techniques in robotics • Deep understanding of estimation in dynamic systems Control system • Probabilistic models • Inference, learning • Hands-on experience World model Actions 10/1/15 Probabilistic Robotics 3 10/1/15 Probabilistic Robotics 4 1
Robotics Yesterday Current Trends in Robotics Robots are moving away from factory floors to • Entertainment, toys • Homes, hotels (personal robotics) • Medical, surgery • Industrial automation (mining, harvesting, warehouses, …) • Hazardous environments (space, underwater, battlefields, …) • Roads 10/1/15 Probabilistic Robotics 5 10/1/15 Probabilistic Robotics 6 Minerva (CMU + Univ. Bonn, 1998) Architecture of the Control System 10/1/15 Probabilistic Robotics 7 10/1/15 Probabilistic Robotics 8 2
RoboCup: RoboCup-99, Stockholm, Sweden Integrated System Research • Focus on addressing all problems at once • Hardware development • Perception • Low level control • High level planning and decision making • Multi robot systems 10/1/15 Probabilistic Robotics 9 10/1/15 Probabilistic Robotics 10 RoboCup: Midsize League RoboCup Small Humanoid League 10/1/15 Probabilistic Robotics 11 10/1/15 Probabilistic Robotics 12 3
RoboCup Rescue DARPA Urban Challenge 2007 10/1/15 Probabilistic Robotics 13 10/1/15 Probabilistic Robotics 14 Google Self-Driving Car Control: BigDog Probabilistic Robotics 10/1/15 15 10/1/15 Probabilistic Robotics 16 4
Cheetah 10/1/15 Probabilistic Robotics 17 10/1/15 Probabilistic Robotics 18 Boston Dynamics Cheetah DARPA Robotics Challenge 2015 10/1/15 Probabilistic Robotics 19 10/1/15 Probabilistic Robotics 20 5
Humanoids: Honda P2 Getting out of Car Honda P2 ‘97 10/1/15 Probabilistic Robotics 21 10/1/15 Probabilistic Robotics 22 Drilling Hole Current Research Trends / Topics • Manipulation of everyday objects • Complex household tasks (cooking, cleaning, …) • Kinect for object detection, 3D mapping, tracking, interaction • Human robot interaction • Machine learning for control, imitation learning, recognition • Deep learning 10/1/15 Probabilistic Robotics 23 10/1/15 Probabilistic Robotics 24 6
Course Outline Sample-based Localization (sonar) Week Content HW / Project #1 Introduction Probabilistic Models / State Estimation #2 Bayesian state estimation / filtering #2 Motion and sensor models, Gaussian processes HW 1: GP modeling Filtering (localization, mapping) #3 / 4 Robot localization: grid, particle filters, EKF, UKF HW2: Filtering #5 / 6 Map building: EKF-SLAM, Fast-SLAM, RGBD Planning / Control #6 / 7 / 8 Path planning, exploration, MDPs, POMDPs Project #9 Reinforcement learning, inverse RL Other Topics #10 Object detection and tracking 10/1/15 Probabilistic Robotics 25 10/1/15 Probabilistic Robotics 26 Graphical Model Representation Mapping with Laser Scanners of Localization Problem 10/1/15 Probabilistic Robotics 27 10/1/15 Probabilistic Robotics 28 7
Mapping with Kinect SLAM: Simultaneous Localization and Mapping u u u 0 1 t-1 ... x x x x 0 1 2 t m z z z 1 2 t 10/1/15 Probabilistic Robotics 29 10/1/15 Probabilistic Robotics 30 Localization and Ball Tracking Structured Estimation • Robot localization l l l z z z k-2 k-1 k Landmark detection r r r k-2 k-1 k Map and robot location u u k-2 k-1 Robot control m m m k-2 k-1 k Ball motion mode • Ball tracking b b b k-2 k-1 k Ball location and velocity b b b z z z • k-2 k-1 k Ball observation 10/1/15 Probabilistic Robotics 31 10/1/15 Probabilistic Robotics 32 8
Gaussian Process Sensor Model Articulated Tracking (42 DOF) for WiFi Signal Strength Mean • Non-parametric regression • GP regression • continuous locations • smooth interpolation • uncertainty estimates Variance Probabilistic 10/1/15 Robotics 10/1/15 33 34 Probabilistic Robotics RL with GP Dynamics Models: Tracking Example PILCO (Probabilistic Inference for Learning Control) Probabilistic 10/1/15 Probabilistic Robotics 35 Robotics 10/1/15 36 9
[Ziebart-Bagnell-etal] Pedestrian Trajectory Prediction Pedestrian Trajectory Prediction ¡ Inverse optimal control: Learn cost function that explains human behavior; use that to estimate goal Probabilistic Robotics 10/1/15 10/1/15 Probabilistic Robotics 38 37 Planning for Manipulation 10/1/15 Probabilistic Robotics 39 10
Recommend
More recommend