Material from the CSE478 mobile robotics course offered at IIIT Hyderabad in Monsoon 2019.
The course introduces the student to fair detail on the basic modules for automating a mobile robot such as state estimation, visual odometry and mapping, planning, and collision avoidance. The course draws upon state of the art practices in probability and statistical methods, optimization techniques and shows how they are dovetailed to a robotics setting. The course has a strong coding component in the form of assignments wherein the student is expected to simulate and implement the algorithms taught in class.
Basic linear algebra, probability theory, and calculus.
Vision Rigid body transformations, Projective geometry, Camera modelling, Camera calibration, Two-view geometry, Stereo, Triangulation, Resection, Visual odometry, Bundle adjustment
State estimation Bayesian filters - Kalman filter, Extended Kalman filter, Localization and Mapping using EKF
Path planning AI-style planning, Kinematics, Randomized planning, Trajectory optimization, Collision avoidance in dynamic environments
Remark Matrix decomposition techniques, such as SVD, QR, RQ, Cholesky, and Linear least squares and Non-linear least squares solvers which will be used throughout the course will be covered as well
-
Hartley, R., & Zisserman, A. (2003). Multiple view geometry in computer vision. Cambridge university press.
-
Thrun, S., Burgard, W., & Fox, D. (2005). Probabilistic robotics. MIT press.
-
Latombe, J. C. (2012). Robot motion planning (Vol. 124). Springer Science & Business Media.
Please refer to the course information document for more resources.
Assignment - 1: Camera modelling and DLT
Assignment - 2: Monocular visual odometry
Assignment - 3: Stereo reconstruction and iterative PnP
Assignment - 4: Localization using EKF
Assignment - 5: Polynomial trajectory planning
Mid-semester exam
End-semester exam
Teams had to read, understand, and present one paper each from the following list of papers. Each paper describe a systems-level project where many different algorithms, some of which where taught during the course, are integrated into one robot system.
Building a Winning Self-Driving Car in Six Months
Design of an Autonomous Racecar: Perception, State Estimation and System Integration
Stereo Vision based indoor/outdoor Navigation for Flying Robots
Camera-Based Navigation of a Low-Cost Quadrocopter
Duckietown: an Open, Inexpensive and Flexible Platform for Autonomy Education and Research
Advanced Autonomy on a Low-Cost Educational Drone Platform
All links were last accessed in Feb 2020.