Course Search, Navigate, and Actuate

"Zoeken, Sturen en Bewegen"

This is the information of year 2012.

The site of the previous year can be found here.
The experiments of the last week are listed on a separate page.

Description

The official description of course baiZSB6 can be found (in Dutch) here. Also a Blackboard portal to this information is available. Blackboard is mainly used for email and grading.

Contents

  • Search Algorithms
    Game playing is an example of type of problems that can easily decomposed in subproblems. For interesting games, like chess, the tree of subproblems grows to fast to be searched exhaustively, so other approaches are necessary. To solve the game we have to find a solution tree regardless of the opponent's replies.
    • MiniMax principle
    • alpha-beta algorithm
    • increasing the effectiveness with advice rules

  • Path planning
    You have had planning algorithms such as A* that work on graphs. So let's try to reformulate the path planning problem as a graph problem. These graphs are somewhat special, it is convenient to see them as discretized spaces because this leads to better implementations. So then we need the notion of configuration space to explain the graph's properties.
    • A* revisited
    • Mapping path planning as graph search
    • Task space and discretized configuration space
    • Kinematics -> connectivity
    • Criteria -> metric
    • Obstacles -> forbidden nodes
    • Examples: robot arm and self-parking car
    • Other approaches of mapping path planning into graphs

  • Trajectory planning
    If you have setpoints, how to make it into a controllable path.

  • Rigid body motion
    • physical rigid bodies as idealization
    • physical space as vector space
    • representing motions using linear algebra (coordinate-free)
    • isometries
    • proof of decomposition theorem: rigid body motion = rotation followed by translation
    • coordinates: vector spaces in the computer
    • rotation matrices: how to design them
    • reference angles: Euler angles
    • homogeneous coordinates

  • Kinematics of linked mechanisms
    • Denavit-Hartenberg notation
    • Forward kinematics
    • Inverse kinematics (briefly)
    • Redundancy and degeneracy (briefly)
    • Differential kinematics

Schedule

This schedule has some correspondance with the official schedule, but in case of doubt use this page as reference.

Week 23

Download Lecturnity Player to listen to lecture, synchronized with the sheets.

Search
(follow_strategy.pl), (advice.pl)
date time type subject location lecturer/assistant
Monday 4/6 9.00-9.15 L0 Course Overview
Lecture pdf, recording lpd
G0.18 Arnoud Visser
Monday 4/6 9.15-11.00 SCR1 Searching through Game Trees - minimax and alpha-beta
Lecture pdf, recording minimax lpd, recording alpha-beta lpd.
G0.18 Arnoud Visser, Elise Koster fig22_3.txt
Monday 4/6 11:00-17:00 SCR2 Searching through Game Trees - Advice Language
Lecture ppt, recording Advice Language for Tic-Tac-Toe lpd
G0.12, G0.23, G0.25 Arnoud Visser, Camiel Verschoor fig22_5.txt
Tuesday 5/6 9.00-11.00 L1 Qualitative Navigation
Lecture (pdf 1.1 Mb), recording (lpd 32 Mb)
Distinctive Place Movie (88 Mb)
Visual Homing Movie (15 Mb)
G2.10 Arnoud Visser
Tuesday 5/6 11.00-13.00 P1 Task 1: Endgames G0.12, G0.23, G0.25 Julian Kooij
Tuesday 5/6 13.00-17.00 ZS Task 1: Endgames G0.12, G0.23, G0.25 No assistance
Wednesday 6/6 11.00-13.00 ZS Task 1: Endgames G0.12, G0.23, G0.25 no assistance, Walibi Day
Thursday 7/6 9.00-11.00 L2 Quantative Navigation
Lecture (pdf 194 Kb), recording lpd, Voronoi graph movies.
P2.10 Arnoud Visser
Thursday 7/6 11.00-13.00 P2 Task 1: Endgames G0.12, G0.23, G0.25 Julian Kooij
Thursday 7/6 13.00-17.00 ZS Task 1: Endgames G0.12, G0.23, G0.25 No assistance
Friday 8/6 11.00-13.00 ZS Task 1: Endgames G0.12, G0.23, G0.25 no assistance, 'Active Leden Weekend'

Week 24

Navigate
date time type subject location lecturer/assistant
Monday 11/6 9.00-11.00 L3 Path Planning algorithms G2.10 Leo Dorst
Monday 11/6 11.00-17:00 Pres demonstration student solutions G0.23, G0.25 Julian Kooij, Arnoud Visser
Monday 11/6 11.00-13:00 P3 Task 2: Path planning module G0.12 Toto van Inge
Tuesday 12/6 9.00-11.00 L4 rotations en homogeneous coordinates C0.110 Leo Dorst
Tuesday 12/6 11.00-13:00 P4 Task 2: Path planning module G0.12, G0.23, G0.25 Toto van Inge
Tuesday 12/6 13.00-17.00 ZS Task 2: Path planning module G0.12, G0.23, G0.25 no assistance
Wednesday 13/6 9.00-11.00 L5 kinematics: Denavit Hartenberg G2.10 Leo Dorst
Wednesday 13/6 11.00-13:00 P5 Task 2: Path planning module G0.12, G0.23, G0.25 Toto van Inge
Wednesday 13/6 13.00-17.00 ZS Task 2: Path planning module G0.12, G0.23, G0.25 no assistance
Thursday 14/6 9.00-11.00 L6 inverse kinematics G2.10 Leo Dorst
Thursday 13/6 11.00-13:00 P6 Task 2: Path planning module G0.12, G0.23, G0.25 Toto van Inge
Thursday 14/6 13.00-17.00 ZS Task 2: Path planning module G0.12, G0.23, G0.25 no assistance
Friday 15/6 9.00-11.00 No Lecture
Friday 15/6 11.00-13:00 P7 Task 2: Path planning module G0.12, G0.23, G0.25 Toto van Inge
Friday 15/6 13.00-17.00 ZS Task 2: Path planning module G0.12, G0.23, G0.25 no assistance

Week 25

Actuate
date time type subject location lecturer/assistant
Monday 13/6 9.00-17.00 TO BE UPDATED
Tuesday 14/6 9.00-17.00 P7 Task 3: Inverse kinematics module G0.18 Julian Kooij
Wednesday 15/6 9.00-17.00 P8 Task 3: Inverse kinematics module G0.18 Julian Kooij
Wednesday 15/6 11.00-13.00 Experiment0 Brainstorming next week G0.18 Arnoud Visser
Thursday 16/6 9.00-17.00 P9 Task 3: Inverse kinematics module G0.18 Julian Kooij
Friday 17/6 9.00-17.00 Pres demonstration and integration Task 2 and 3 G0.18 Julian Kooij

Week 26

Go, where no one has gone before.
date time type subject location lecturer/assistant
Monday 20/6 9.00-12.00 Experiment1 Kick-Off G0.18 Arnoud Visser
Monday 20/6 12.00-17.00 ZS G0.18 no assistance
Tuesday 21/6 9.00-17.00 ZS G0.18 no assistance
Wednesday 22/6 9.00-12.00 Experiment2 Mid-Term G0.18 Arnoud Visser
Wednesday 23/6 12.00-17.00 ZS G0.18 no assistance
Thursday 23/6 9.00-17.00 ZS G0.18 no assistance
Friday 24/6 9:00-17.00 Experiment3 Demonstration and Documentation G0.18 Arnoud Visser
Friday 24/6 from 17.00 Experiment4 Barbecue near cantine VIA

It is not the result that counts, but your summery of your survey. Document your progress, experiments and decisions in a LabBook.

With a working system, and the acquired knowledge, you can explore new possibilities.

Here are the surveys of the previous years:

Here are some suggestions:
  • Path-planning for a Hemisson-robot
  • Talking mouth for a Aibo-robot
  • Maze navigation with a Nao-robot
  • Looking to a talking person with a Nao-robot.
  • Extend the checkmate problem to more complex situations
  • Refine the visualisation of the Virtual robot.
  • Creating a gamepad interface for a virtual Aibo (Visual Basic)
  • WiiBot RTX UMI
  • Solve chess endgame with Monte Carlo tree search (MCTS)
It is recommanded to work in groups of three students.

You will be evaluated on your LabBook at the end of the week.

Evaluation

This year the course was overall evaluated by the participants with a 7.7 .

Literature


For the implementation in prolog we will look at chapter 24 of Prolog Programming for Artificial Intelligence by Ivan Bratko. The companion website of the 4th edition is not yet available, the companion website of the 3th edition contains several student resources.
This book was explored until chapter 13 in the previous course Logic Programming and Search Techniques.


A short introduction will be given based on the second part of Introduction to AI Robotics by Robin Murphy: Navigation.
Part I book was in the previous years explored in the course Reactive Behaviours. Now this course is replaced the more advanced course Probabibilistic Robotics / Autonomous Mobile Robots. Murphy's book is not longer required, and should not be bought. The two chapters are available electronically (Chapter 9 and Chapter 10).

Further we use the syllabus 'An Introduction to Robotics' by Leo Dorst and a lab manual. The syllabus available from the Dikatenverkoop (check the opening times at the VIA-site).


Inheritance

In the old days, when Bachelors were not schooled at Dutch Universities, a different course was given with another focus. Still, much can be learned from the course 'Robotica'.


Last updated July 3, 2012

o This web-page and the list of participants to this course is maintained by Arnoud Visser (arnoud@science.uva.nl)
Faculty of Science
University of Amsterdam

visitors in arnoud@science.uva.nl