INFT840
Advanced Topics in Robotics and Computer Vision
Time/Location: Wednesday 4:30- 7:10 p.m., Robinson B-220
Instructor: Jana Kosecka
Office Hours: tba

Course Outline:

  • Introduction, motivation, overview (geometry, statistics, optimization, algorithmic issues).
  • Projection and camera models (extrinsic and intrinsic parameters, perspective, spherical, orthographic, paraperspective models)
  • Rigid body motion (properties, characterizations, representations).
  • Kinematics
  • Structure and motion estimation problem (formulation discrete and differential case)
  • Measurement stage (computing derivatives, optical flow, correspondences)
  • Epipolar geometry and motion estimation (2-3) (linear techniques, discrete and differential case)
  • Nonlinear optimization techniques (overview of constrained and unconstrained optimization techniques, differential geometric approach)
  • Sensitivity analysis of motion estimation (ambiguities, sensitivity, choice of the proper error metric)
  • Structure recovery and Multi-view geometry (different structure and camera models, two frame and multiframe frame techniques (discrete and differential case), classical group invariants)
  • Epipolar geometry (uncalibrated case, fundamental matrix, translation and rotation case, conditions and ambiguities for calibration, structure and motion recovery)
  • Camera calibration and self-calibration (theory and algorithms)
  • Recursive techniques for motion estimation
  • Dealing with noisy measurements and multiple motions

  • Calibration with respect to motion and calibration subgroups, reprojection
  • Kinematics of articulated bodies (product of exponentials, human motion tracking, articulated motion)
  • Visual servoing - kinematic and dynamic formulation (Lagrange equations of motion)
  • Feedback control using measurements in the image plane.
  • Visual servoing - Case studies of docking and aircraft landing.
  • Overview of motion planning techniques.
  • Visually guided navigation in the mobile robot context (obstacle avoidance and motion planning techniques)

    Grading:
    Homeworks: 30%
    Class Participation: %30
    Final project: 40%
    Prerequisites: linear algebra, calculus