EECS 598-2: Algorithms for Robotics (W2009)

General Information

Materials

Problem Sets


Announcements

3/11: Class is canceled today. The chapter on bug algorithms is available here: here (look under errata).

Please look through these slides for an idea of what we planned to talk about today. We'll touch on the big points next time, but this will be more efficient if the ideas are somewhat familiar.

Syllabus

Course Objective

This course will present and critically examine contemporary algorithms for robot perception (using a variety of modalities), state estimation, mapping, and path planning. Programming exercises and a project will give students the opportunity to try algorithms themselves and to propose improvements. Successful students will finish the course with a first-hand knowledge of the essential algorithms, their advantages and disadvantages, and an understanding of practical implementation issues.

Assignments

There will be four homework assignments, due at intervals of two weeks, followed by a project period lasting six weeks. Assignments will generally involve implementing algorithms and evaluating their performance. The project will involve implementing a more complicated algorithm, or implementing a system comprised of several algorithms, or exploring improvements to existing algorithms. Students will write a project proposal, a mid-way progress report, and present their final project.

Exams

There will be no exams.

Grading

Homework assignments: 64% (16% each)
Project: 36% (6% proposal, 10% mid-way report, 20% final presentation and report

Collaboration

Collaboration is encouraged, but all collaboration must be reported. Students may collaborate on homeworks in groups of up to three, including submitting solutions that were jointly developed. Any joint collaboration must be done with the students physically working over the same computer or notebook at the same time in a "pair programming" style: dividing up problems and delegating the work is not acceptable. All group members are fully responsible for every part of their homework. Groups are free to talk with other groups, but may not share any non-trivial solutions or code.

Projects may be done individually or in small groups, though the scope of the project must reflect the resources available.

Schedule

DateLectureHomework
Wed Jan 7L01. Course OverviewPS1 Out 
Mon Jan 12L02. Probability Review; Multi-Gaussian Distributions 
Wed Jan 14L03. Rigid-Body Transformations. Maximum Likelihood SLAM 
Wed Jan 21L04. Extended Kalman Filter 
Mon Jan 26L05. Extended Kalman Filter (continued) 
Wed Jan 28L06. Particle FiltersPS1 Due. PS2 Out 
Mon Feb 2L07. Particle Filters (continued) 
Wed Feb 4L08. Data Association 
Mon Feb 9L09. Topological Mapping 
Wed Feb 11L10. LIDAR feature extractionPS2 Due. PS3 Out 
Mon Feb 16L11. LIDAR scan matching 
Wed Feb 18L12. Camera basics; calibration 
Mon Mar 2L13. Camera features 
Wed Mar 4L14. Kinematics, Inverse-Kinematics, Motion control lawsPS3 Due. PS4 Out 
Mon Mar 9L15. Deterministic Motion Planning 
Wed Mar 11L16. Non-deterministic Planning (CANCELED) 
Mon Mar 16L17. Non-deterministic Planning 
Wed Mar 18L18. ControlProject Proposals Due. 
Mon Mar 23L19. System design (real time issues, communciation)PS4 Due 
Wed Mar 25L20. Fast linear algebra 
Mon Mar 30L21. Non-Linear SLAM revisited (SqrtSAM) 
Wed Apr 1L22. Gauss-Seidel, SGD SLAM 
Mon Apr 6L23. sensors (low-cost, radar, velodyne, stereo, exotic)Project Status Report Due 
Wed Apr 8L24. Robust Place Recognition 
Mon Apr 13L25. Case study: DARPA Grand Challenges 
Wed Apr 15L26. Final Project Presentations 
Mon Apr 20L27. Final Project Presentations 
Tue Apr 21(no lecture)Project Reports Due 
Topic color coding
Miscellaneous topics
State Estimation and Mapping
Sensing
Path Planning

Readings


EECS 598-2: Algorithms for Robotics, Winter 2009.