Information Processing in Robotics

Fall Semester 2010 / 2011

Useful links

• Setting up ROS (install diamondback and not electric): Summary document and official page
• ROS dependencies for the simulation environment
• ROS tutorials

Classes

• Class 1 Introduction to Information Processing
• Class 2 Intro to Learning and probabilistic Reasoning
• Class 3 Graphical Models and Hidden Markov Models
• Class 4 Online estimation (KF, EKF), application to localisation and mapping
• Class 5 Regression
• Class 6 Kernel Methods for regression, Gaussian Processes
• Class 7 Kernel Methods for classification, Support Vector Machine
• Class 8 Iterative Closest Point
• Class 9 Sampling Methods, application to localisation
• Class 10 Mean Shift and Clustering approaches
• Class 11 Expectation Maximisation
• Class 12 Robust estimation using Hough Transform and RANSAC
• Class 13 Classification for computer vision: PCA and AdaBoost
• Class 14 Robotic Challenge

Exercises

• Class 1 ROS Package: Face Room. Untar into your ROS_PACKAGE_PATH, check the dependencies and make. Use the launch file to start the simulator.
• Class 2: Exercise sheet, solution, code
• Class 3: Exercise sheet, code, solution,
• Class 4: Exercise sheet, office_slam package, code, solution,
• Class 5: Exercise sheet, code, solution,
• Class 6: Exercise sheet, code skeleton, solution, final code,
• Class 7: Exercise sheet, code skeleton, solution, final code,
• Class 8: Exercise sheet, office_icp package, code skeleton, final code,
• Class 9: Instructions, ros package to be completed, final code
• Class 10+11: Instructions, Code
• Class 12: Instructions and Code Base
• Class 13+14: Robotic Challenge

Oral Exam Questions

The first part of the IPR examination will consist in a discussion (~ 15min) about a particular robotic systems selected in the list below. The candidate is expected to discuss how tools presented in the IPR class could be used to implement the selected application. Examples of topic to discuss include: perception processing, control, object/event detection, object/event classification, learning opportunities, localisation, uncertainties... This list is not exhaustive, and some of the topics may not apply to a particular system.

• Application Questions (read this one first).

The second part of the examination (~15min) will focus on specific technical questions drawn out of this list.

Material from previous years

• Fall semester 2008 / 2009
• Fall semester 2009 / 2010
• Fall semester 2010 / 2011