Professor: Marek A. Perkowski,
Electrical and Computer Engineering.
Intelligent Robotics Laboratory
EE 478/578 Intelligent Robotics I (4).
This is material for school year 2005.
TEXTBOOKS:
- Fred G. Martin, "Robotic Explorations. A Hands-On Introduction to Engineering," Prentice Hall.
- Stuart Russell and Peter Norvig, "Artificial Intelligence. A Modern Approach." Prentice Hall.
- George F. Luger and William A. Stubblefield,
"Artificial Intelligence. Structures and Strategies for Complex Problem Solving,"
Second Edition. Benjamin/Cummings Publ. Co., 1992.
- Marek Perkowski, Anas Al-Rabadi, Alan Mishchenko,
``Evolvable Robots, Universal Decision Diagrams, Quantum Logic, Intelligent
Animated Puppets, and Other Machine Learning Projects for Inquisitive Minds''.
(Textbook in preparation). first part in postscript
NOTES ON TEXTBOOKS USE:
- The main class textbook is Martin and Russell/Norvig (RS).
- You can also use the book by Luger/Stubblefield (LS).
- Our book has ONLY auxiliary material and does not replace RN or LS.
- All lecture slides and auxiliary notes are available on this WWW Page for class use.
Read them systematically.
GOALS:
-
The course introduces the main software ideas used in intelligent robotics
and how some of them are realized in specialized hardware processors.
-
Students learn about predicate logic, logic programming, state-space methods,
heuristic search, and functional programming.
-
The methods of formulating and solving the robotic problems
in various models are stressed.
-
Understanding of software/hardware tradeoffs is emphasized.
-
Every student participates in building a robot.
Projects are assigned to groups or individual students.
Students completing this course have a good working knowledge of
fundamentals of intelligent robotics and are prepared for second
class in the sequence.
It is also important that
students learn practical robot-building skills, materials, supply sources and after the class
are not afraid of using their hands in things like: cutting aluminum, PCB design, soldering, or debugging.
PREREQUISITES BY TOPIC:
Background in Pascal, or C programming and basic digital
design from EE 485/585 or EE 425/525.
TOPICS:
1. Introduction to Robotics and Artificial Intelligence.
2. Programming in LISP.
3. State Space Search.
4. Heuristic Search, problem representation.
5. Predicate Calculus and Resolution Method.
6. Robot architectures.
7. Reasoning by analogy and induction.
8. Computer architectures for logic programming and Artificial Intelligence.
Role of FPGAs, evolvable hardware and reconfigurability.
9. Introduction to machine learning and computer vision.
10. Rule-based systems and expert systems.
11. Fuzzy logic and applications.
PROJECTS FOR YEAR 2005:
1. Talking and interacting humanoid robot (latex skin version): Facial gestures animation.
2. Talking and interacting humanoid robot:
(both latex skin and wood sculpture versions). Text and jaw movement synchronization.
3. Human Gesture and emotion recognition based on Computer Vision with several cameras.
4. Automated robot theatre: scripts and dialogs. User interaction.
5. Automated robot theatre: mechanical and electrical
design of korean mask Hahoe robots. Animation of dance and songs. Control of large servos (software and hardware).
6. Evolutionary Quantum Breitenberg Vehicles:
This is a small mobile robot (it already exists)
that learns in quantum logic from examples. Implemented as evolvable hardware using Altera FPGA.
7. Hexor robot - hunting for soda cans with vision.
8. Radio-control and robot vision for walking robot Robosapiens.
9. A camera controlled pair of walking japanese robots.
PREVIOUS LABORATORY PROJECTS (EXAMPLES):
-
The projects varied from year to year and included:
designing small mobile and stationary robots,
interfacing and programming robots,
writing programs in LISP, PROLOG, and C/C++;
using image processing packages;
designing and simulating specialized computer architectures with FPGAs,
-
The program examples include: state-space search, control of mobile
robots in labirynths, obstacle avoidance, solving puzzles,
graph-theory problems, using graph models in computer vision and planning,
reasoning by analogy, rule-based systems, image processing (Hough Transform, filters,
histograms),
natural language communication with mobile robot in PCAT building environment.
-
The FPGA-VHDL designs included the following architectures: computer for predicate calculus,
Hough Transform, Fast Fourier Transform, sonar controller, motor controllers.
COMPUTER USAGE:
-
Students use the departmental network of Sun workstations and
PC-based computers to solve various problems.
-
Several projects are also related to PSUBOT - a PC-based wheelchair robot
designed in our department.
ESTIMATED CONTENT:
-
Engineering Science: 2 credits or 50%.
-
Engineering Design: 2 credits or 50%.
DETAILED TOPICS THAT WILL BE DISCUSSED IN 2004.
Order of presentation will depend on class projects and students' choices of projects.
- Introduction to Robotics and Artificial Intelligence.
- Overview of robotics in the world; from robotic toys to military and medical robots.
- Robot kits and technologies that we use in class.
- Lego and Dacta
- Robotix
- Lynxmotion
- Other kits
- Mechanical prototyping
- Motors
- Sensors
- Electrical prototyping: Field Programmable Gate Arrays, Field Programmable Analog Arrays, EPLDs.
- Computer interfacing.
- Software for motion control
- Overview of basic problems and
past research in Artificial Intelligence, Machine Learning, and Intelligent Robotics.
- Presentation and discussion of class projects for 2001.
- Programming in LISP.
- Basic LISP functions. Lists, trees, graphs.
- Recursion
- Property lists.
- Robot control.
- Advanced Data Structures.
- Robot Vision.
- Natural Language Processing, conversation and story telling.
- Predicate Calculus and Resolution Method.
- Introduction to Prolog.
- Introduction to Constraints Programming.
- Automatic Theorem Proving.
- Hardware Realizations of Prolog and Theorem Proving: unification and resolution in hardware.
- Robotics Problems as constraints problems.
- State Space Search.
- Depth-First and Breadth-First Search algorithms and applications.
- A* Search and advanced search in OR-trees.
- Search in AND/OR-trees and DAGs.
- Path planning and navigation as search.
- Heuristic Search, problem representation.
- Rule-Based systems.
- Representation of robotics problems as heuristic search problems.
- Genetic Algorithm and Genetic Programming.
- Artificial Neural Nets.
- Reasoning by analogy and induction.
- Reasoning by analogy
- Reasoning by induction
- Abductive reasoning.
- Generalization.
- Computer architectures for Artificial Intelligence.
- Lisp Machines.
- Cube Calculus Machine.
- Satisfiability Machines.
- Search Engines.
- Military Data base machines based on Predicate Calculus.
- Associative Memory and Associative Processors.
- Introduction to machine learning and computer vision.
- Formulation of basic problems related to robotics.
- Rule-based systems and expert systems.
- Formulation of basic problems related to robotics.
- Fuzzy logic and applications.
- Fuzzy logic as extension of binary and multivalued logic. Transformations and minimization.
- Designing of Fuzzy logic controllers.
- Decomposition of Fuzzy functions and relations.
- Fuzzy logic VLSI chips and applications.
- Discussion of class projects.
PAST PROJECTS.
- PEOPLEBOT - sonar and radio controlled mobile robot. This is purely software project, natural
language communication with humans. Robot entertainer for PSU meetings.
- PSUBOT - sonar and vision controlled wheelchair for severely handicapped.
A new generation of assistive robot.
- MUVAL - a stationary robot with 4 hands and 2 legs. Reasoning in Multiple-Valued Logic. Image Processing.
- OREGON CYBER THEATRE - a variety of stationary, wheeled and walking robots-pupetts.
Hexapods, quadrupeds and bipeds. Improve mechanics, add sensors, and improve software.
Radio control. Interaction.
- ELECTRIC HORSE - walking robot for Rose Parade.
- MOLLY - an interactive pupett-girl, C++ controlled, based on inexpensive servos.
- HERO - the oldest robot in our lab, born in 1981, still operational but needs some fix and better software.
Add talking head.
- TALKING HEADS - several talking heads for speach recognition/synthesis and control of
emotional behaviors.
- HALLOWEEN SKELETON - talking skeleton does some gymnastics, programmed using object-oriented
PIC microcontroller.
- ROBIX - big plastic humanoid type, on wheels,
built from commercial Robix kit, needs head replacement and smarter control.
- RHINO - classical robotic arm, needs to add LISP control to the current ROBOTALK language.
- HEXAPOD CENTAUR - centaur with six legs, needs body and head.
- A LEG - pneumatic control of a horse leg model with 9 degrees of freedom.
- A CRANE - design a complete control of a student-built crane-arm model.
- HOME DEPOT ROBOT - robot built from inexpensive home-building materials. Classical robotics look
such as RUR or cinema robots, will be built for communication and entertainment.
- PIPE ROBOT - SARCOS-like inexpensive robot built from pipes. Emphasis of this project is on
intelligent movement control of heavy body components.