Mobile Robots Summery. Autonomous Mobile Robots

Transcription

1 Mobile Robots Summery Roland Siegwart Mike Bosse, Marco Hutter, Martin Rufli, Davide Scaramuzza, (Margarita Chli, Paul Furgale) Mobile Robots Summery 1

2 Introduction probabilistic map-based localization knowledge, data base Localization Map Building position global map Cognition Path Planning mission commands environment model local map path Perception Information Extraction raw data Sensing see-think-act Path Execution actuator commands Acting Motion Control Real World Environment Mobile Robots Summery 2

3 Legged Robots and Kinematics Types and application of legged systems Number of legs Analogy to nature Static and dynamic stability Locomotion control Basics of rigid body kinematics Translation, rotations, and homogeneous transformation Translational and angular velocities Rigid body kinematics formulation Vector differentiation in moving coordinate systems Application of kinematics in robotics Generalized coordinates and Jacobians Forward and inverse (differential) kinematics Redundancy and singularity Floating base systems and contact constraints Mobile Robots Summery 3

4 Wheeled Locomotion Wheeled types and arrangements Kinematics Constraints imposed by wheels Forward or inverse differential kinematics Analysis of the differential kinematics equations the degree of maneuverability = degree of mobility + degree of steerability Mobile Robots Summery 4

5 Commputer Vision Projective Geometry Perspective projection Intrinsic and extrinsic parameters Stereo vision Correspondence search Rectification Disparity map Z P Perspective Projection Matrix X w u Yw v K RT Z w 1 1 bf u u l r Disparity Structure from motion Epipolar geometry Epipolar constraint Essential matrix 8-point algorithm,?? epipolar line C l epipolar plane epipolar line C r Mobile Robots Summery 5

6 Image Saliency image filtering & place recognition Image Filtering: Correlation vs. Convolution Use in template matching, smoothing & taking the derivate of an image Image filtering for Edge Detection Building and using the visual vocabulary for Place Recognition 2N+1 2N+1 Point Features: Harris, SIFT, FAST, BRIEF, BRISK & their characteristics e.g. scale/rotation invariance, computational time Examples of Visual Words Mobile Robots Summery 6

7 Line Fitting algorithms & error propagation The Error Propagation Law How uncertainties propagate through a function. Line Fitting algorithms for image/laser point clouds Split-and-merge, RANSAC, Hough Transform,.. How they work & their relative characteristics and applications Courtesy of ETH - Mobile Robots Summery 7

8 Localization where am I? SEE: The robot queries its sensors finds itself next to a pillar ACT: Robot moves one meter forward motion estimated by wheel encoders accumulation of uncertainty SEE: The robot queries its sensors again finds itself next to a pillar Belief update (information fusion) Mobile Robots Summery 8

9 SLAM approaches & current challenges What is SLAM and how does it work? The graphical representation SLAM & the approaches to solve it: Full graph optimization Filtering Keyframe-based A B C Popular techniques & how they work: EKF SLAM via MonoSLAM [Davison et al. 2007] SLAM today & Challenges Mobile Robots Summery 9

10 Motion Planning the planning problem Goal Mobile Robots Summery 10

11 Motion Planning hierarchical decomposition & approaches 1. Local collision avoidance Dynamic Window Approach (Reciprocal) Velocitiy Obstaces Local potential fields v 2. Global planning Harmonic potential fields Graph search (BF, Dijkstra, A*) Randomized tree search (RRT) A 1 D E B 3 C Mobile Robots Summery 11

12 Beyond Mobility PR2 robot from Willow Garage Fold towels Courtesy of Clean-up Mobile Robots Summery 12

13 Exam Oral, 30 minutes, August Questions selected by examiners Application Basics 1 Basics 2 Questions given beforehand (Webpage, sent to all participants) All terrain demining in unstructured environments Wheel kinematic constraints of the 5 wheel types, pro/cons of wheel types Odometric position estimation and error model for a differential drive robot and their use in Markov and EKF localization 12 - Summery 14