Winter Term 13/14
The course will introduce basic and advanced concepts and methods in robotics, machine learning, artificial intelligence and human-robot interaction. While the methods are of general interest for students interested in advanced robotics, we will also discuss techniques that consider the "human in the loop" – a growing field of activity in robotics with new exciting applications. The challenge is to develop key technologies that enable robots to better live, work and interact with humans. In addition to their theoretical treatment, the methods will be exemplified with programming exercises. The course outline is as follows:
- Matlab/Octave introduction
- Probability refresher, common distributions
- Probabilistic reasoning, Bayes networks, Markov chains
- Perception of Humans
- Supervised learning: naive Bayes, logistic regression, SVM, AdaBoost, k-NN, cross-validation
- Unsupervised learning: hierarchical clustering, k-means, GMM and EM
- Temporal Reasoning: Hidden Markov Models, LDS and Kalman filtering: representation and inference
- Tracking and data association: NN, GNN, PDAF, JPDAF, MHT
- Planning among Humans
- Robot motion planning: combinatorial techniques, PRM, RRT, potential fields, A*, Theta*, obstacle avoidance
- From plans to policies: Markov Decision Processes (MDP)
- Interaction with Humans
- Introduction to Human-Robot Interaction
The course is accompanied by programming exercises in Matlab. We assume that students have a laptop with Matlab installed (download and install instructions can be found here). Octave, the open-source alternative to Matlab, can also be used (download for free from www.octave.org).
- Lectures: Monday 14h-16h, Room SR 00-034, Building 051
- Exercises: Wednesday 12h-14h, Room SR 00-031, Building 051
- Exams: March 14/26, 2014, Building 074, Office 015
- Language: English
- Recordings: I want you to participate :-) so no recordings
- Solving and submitting the exercise sheets is not mandatory to be admitted to the final exam. Of course, it is highly recommended for the sake of a successful exam.
- There are no bonus points, no exam admission requirements.
- In general, assignments will be published on Monday and have to be submitted the following Monday before class.
- Submit programming exercises via email to the address indicated on the respective sheet.
- Exercise 1, Matlab/Octave basics
- Exercise 2, Introduction to probability
- Exercise 3, Basics of probabilistic reasoning
- Exercise 4, Naive Bayes classifier
- Exercise 5, Support vector machines
- Exercise 6, k-NN, cross-validation
- Exercise 7, k-Means clustering
- Exercise 8, Hidden Markov model
- Exercise 9, Kalman filter
- Exercise 10, Tracking and data association
- Exercise 11, Rapidly-exploring random trees
- Exercise 12, Q&A session
Slides and Additional Material
||Basics of Probabilistic Reasoning
||Supervised Learning 1/3:
Basics, Naive Bayes, logistic regression
||Supervised Learning 2/3:
Support Vector Machines
||Supervised Learning 3/3:
AdaBoost, k-NN, cross-validation
AHC, k-means, GMMs and EM
- K-means and K-medoids Applet by E.M. Mirkes. Very instructive!
- EM for Mixture Models Applet by I. Dinov. Very instructive!
- Introduction to Data Mining by P.-N. Tan, M. Steinbach, V. Kumar, Addison-Wesley, 2005
- Pattern Recognition by S. Theodoridis, K. Koutroumbas, 4th ed., Elsevier, 2009
- Data clustering: 50 years beyond K-means by A.K. Jain, Pat. Rec. Letters, 31(8), 2010
||Temporal Reasoning 1/3:
||Temporal Reasoning 2/3:
LDS, Kalman filter
||Temporal Reasoning 3/3:
Tracking and Data Association
||Robot Motion Planning:
Basics, combinatorial and sampling-based methods
- Robot Motion Planning by J.C. Latombe, Kluwer Academic Publishers, 1991
- Planning Algorithms by S. LaValle, Cambridge University Press, 2006
- Principles of Robot Motion: Theory, Algorithms, and Implementations by H. Choset, K.M. Lynch, S. Hutchinson, G. Kantor, W. Burgard, L.E. Kavraki, S. Thrun, MIT Press, 2005
- Chapters 5 and 6 in Handbook of Robotics ed. by B. Siciliano, O. Khatib, Springer, 2008
- Configuration Space Applet by K. Goldberg, E. Lee, J. Wiegley. Very instructive!