Adaptive Robots
mailto:hhl@daimi.aau.dk.
students, groups, and calender
Mini-symposium on Robotics in Biology:
On Friday the 21st of November,
there will be a mini-symposium on robotics in biology to be held at the Natural
History Museum in Aarhus. The speakers are Barbara Webb from Univ. of
Nottingham, Holk Cruse from Univ. of Bielefeld, Christian Freksa from Univ. of
Hamburg, and Henrik Hautop Lund from CIT, Univ. of Aarhus. Have a look at the schedule for mini-symposium.
As an interesting supplement to our discussion about embodied intelligence
during the last two lectures, you may notice that there is a symposium on Embodied Cognition and
AI at 14th European Meeting on Cybernetics and Systems Research, April
14-17, 1998. In the call for contributions, the chairs, Maja Mataric and Erich
Prem, state that "Embodied artificial intelligence (EAI) has recently received
enormous interest both from scientists and the broad public. Embodied artificial
intelligence proposes a new approach to the study of human intelligence and the
construction of intelligent artefacts. Its central belief is that the body is
the logical and physical source of intelligence and that in order to understand
intelligent behavior, dynamic bodily interaction must be understood first."
Further, there has been a very recentspecial issue of
Cybernetics and Systems on Embodied AI. So, while most of our literature in
the course on this issue dates back to mid/late eighties, there is still a
lively discussion going on about the role of Embodied AI.
Multimedie Event:
We will present some of our robots at the Multimedie Event '97, which is held in Aarhus
Raadhus, 19-20 September 1997. Our show will feature a team of garbage
collecting miniature robots.
Papers to read
1. What are Adaptive Robots?
- C. Malcolm, T. Smithers, J. Hallam "An Emerging Paradigm in Robot
Architecture"
- R. Pfeifer "Building "Fungus Eaters": Design Principles of Autonomous
Agents"
- R. Brooks "Artificial Life and Real Robots"
2. Behavior-Based
Systems
- C. Malcolm -- review of Brooks' robotics approach from 1988
- R. Brooks "A Robust Layered Control System For A Mobile Robot"
- P. Maes "Behavior-Based Artificial Intelligence"
- C. Malcolm, T. Smithers "Programming Assembly Robots in terms of Task
Achieving Behavioural Modules: First Experimental Results"
3. Control
Theory
- T. Smithers -- notes on Control and Control Strategies
- J. Hallam "Notes on Control Theory"
4. Evolutionary Robotics
- H. H. Lund, J. Hallam "Sufficient Neurocontrollers can be Surprisingly
Simple"
- I. Harvey, P. Husbands, D. Cliff "Issues in Evolutionary Robotics"
- R. Brooks "Evolutionary Robotics; Where From and Where To"
5.
Simulation or Real Robots?
- O. Miglino, H. H. Lund, S. Nolfi "Evolving Mobile Robots in Simulated and
Real Environments"
- M. Mataric, D. Cliff "Challenges in Evolving Controllers for Physical
Robots"
- D. Floreano, F. Mondada "Evolution of Homing Navigation in a Real Mobile
Robot"
6. Co-evolution of Robots
- D. Floreano "Reducing Human Design and Increasing Adaptability in
Evolutionary Robotics"
7. Towards Complex Behaviours (Evolvable
Behavior-Based Systems)
- W-P. Lee, J. Hallam, H. H. Lund "Learning Complex Robot Behaviours by
Evolutionary Approaches"
- S. Nolfi "Adaptation as a more powerful tool than decomposition and
integration"
- S. Nolfi "Using emergent modularity to develop control systems for mobile
robots"
8. Learning and Evolution
- S. Nolfi, J. Elman, D. Parisi "Learning and evolution in neural networks"
- D. Floreano, F. Mondada "Evolution of Plastic Neurocontrollers for
Situated Agents"
- S. Nolfi "Evolving non-Trivial Behaviors on Real Robots: a garbage
collecting robot"
9. Evolvable Hardware in Robotics
- A. Thompson "Artificial Evolution in the Physical World"
- D. Keymeulen, Kenji Konaka, Masaya Iwata, Yasuo Kuniyoshi, T. Higuchi
"Robot Learning using Gate-Level Evolvable Hardware"
10. Evolving the
Robot's Physical Structure
- H. H. Lund, J. Hallam, W-P. Lee "Evolving Robot Morphology"
- P. Funes, J. Pollack "Computer Evolution of Buildable Objects"
11.
Robots as Biological Models
- H. H. Lund, B. Webb, J. Hallam "A Robot Attracted to the Cricket Species
Gryllus bimaculatus
- D. Lambrinos, M. Maris, H. Kobayashi, T. Labhart, R. Pfeifer, R. Wehner
"An autonomous agent navigating with a polarized light compass"
- N. Franceschini, J. M. Pichon, C. Blanes "From insect vision to robot
vision"
12. Learning
- L. Steels, P. Vogt "Grounding adaptive language games in robotic agents"
- J. Tani "Dynamical Systems Approach to Represent Cognition of Robots: A
View of the Internal Observer"
- K. Balakrishnan, V. Honavar "Spatial Learning for Robot Localization"
13. Reinforcement Learning
- D. Touretzky, L. Saksida "Skinnerbots"
- E. Uchibe, M. Asada, K. Hosoda "Vision Based State Space Construction for
Learning Mobile Robots in Multi Agent Environments"
14. Multi-Robots
- T. Fukuda, G. Iritani "Evolution and Self-Organizing Robots - Artificial
Life and Robotics"
- M. Mataric "Reinforcement Learning in the Multi-Robot Domain"
- O. Holland, C. Melhuish, S. Hoddell "Chorusing and Controlled Clustering
for Minimal Mobile Agents"
15. Vision
- I. Horswill "Visual routines and visual search: a real-time implementation
and an automata-theoretic analysis"
- M. Franz, B. Scholkopf, H. Bulthoff "Homing by Parameterized Scene
Matching"
Book:
- V. Braitenberg "Vehicles: Experiments in Synthetic Psychology", MIT Press,
MA, 1984. ISBN 0-262-52112-1 (paper), 152 pages.
Supplementary reading:
- J. Jones, A. Flynn "Mobile Robots: Inspiration to Implementation", AK
Peters, 1993. ISBN 1-56881-011-3 hardcover, 368 pp.
This is a good book if
you want to get more knowledge about the functionality of sensors, motors, and
how to build your own mobile robot.
- Dorigo M. & M. Colombetti (1997). Robot Shaping: An Experiment in
Behavior Engineering. MIT Press/Bradford Books, in press.
Authors:
(some)
Groups:
- VUB AI Lab
- Mobile Robots Group,
Dept of AI, University of Edinburgh
- Research Group on
Artificial Life, National Research Council, Rome
- LAMI, EPFL, Lausanne
- AILab, University of
Zurich
- AnimatLab, Ecole Normale
Superieure, Paris
- Evolutionary and
Adaptive Systems at COGS, University of Sussex
- Applied AI Systems, Inc., Canada and
Japan
- Sony CSL, Inc., Tokyo
- Miura-Shimoyama
Lab, University of Tokyo
- Humanoid
Interaction Lab, Tsukuba
- The
Humanoid Project, Waseda University [¿Discontinuado?]
- The Interaction Lab,
Brandeis University
- MIT, Artificial Intelligence Lab
- Learning Robots
Lab, Carnegie Mellon University
- Machine Learning
Section, Naval Research Laboratory, Washington, DC
- Robotics Research, University
of Southern California
- Microrobots Group,
MIT, U.S.A.
Khepera robot: the meaning.
You keep wondering what the meaning of the
word Khepera is. Well, here is an explanation.
Khepera robot communication:
From kermit, when the robot is in mode 3,
you should set the communication as follows:
set line /dev/ttya
set speed 38400
set terminal bytesize 8
set
duplex half
con
You can then type in commands from the protocol such as D,2,2
Later, after having made tests with the Khepera robot via kermit, you would
probably like to make complete control programs. In the C programming language,
you can use the communication procedures in serial.tar.gz. In this
tar file, you'll also find Rene's example program that allows the Khepera robot
to turn around in steps of 90 degrees and move backward in steps of 1.0cm.
Khepera Braitenberg program:
Jørgen, Jesper and Henrik have made a nice
obstacle avoidance behaviour for the Khepera robot based on Braitenberg's
principles. The program is also a nice example of how to make the serial
communication with the Khepera robot and how to write the control program in C.
You can download braitenberg.tar.gz.
Khepera Evolutionary Robotics simulator:
This simulator allows you to
evolve neural network control systems for the Khepera robot in simulation and
then run them to the real robot. The simulator also allows evolution directly
(and eventually entirely) on the physical robot. It was made by S. Nolfi, O.
Miglino and H. H. Lund. You can download evolrob.tar.gz
.
You can get more info about the evolutionary robotics simulator in the guide
The compilation has been tested on SUN and SGI (new updated makefile 6/10/97
!)
Remember to use something like nice or hugin
when you run the simulator on other people's machines.
You can untar the files using gunzip and then tar xvf
evolrob.tar (or eventually tar zxvf evolrob.tar.gz).
Run autunomously without cable connection:
After having made a lot of
experiments with the robot with cable connection (in order to get knowledge
about the sensor and motor responses) it is time to look at how to download
program to the robot and have it to run autonomously without cable connection.
This demands cross-compilation of your programs, so that they can run on
Khepera's Motorola 68331. You can still write the programs in C, but the
protocol for communication with the Khepera motors, sensors, timer, etc. is
different. You can download
download.tar.gz that includes sample programs and a makefile that allows you
to cross-compile the programs for the Khepera. You might have to change paths in
the Makefile to suit your local settings (where you have placed the GNU C
cross-compiler).
You can also download
wall.tar.gz which is a small example program that allows the robot to move
forward until it meets a wall and then adjust itself to stand straight in front
of the wall. You can use this as a basis for developing more complex behaviours.
The robot with this small control program will also follow a moving object, as
long as the object is in the vicinity of the robot.
Communication to additional Khepera modules:
In programs downloaded to
the Khepera robot, there is a way to use the same commands as in the protocol
used with serial communication. You can download camera.tar.gz
that gives you an example of how to do this, in this case with the K213 vision
module.
Khepera Potato Holiday Contest:
On Thursday 16/10/97, we will have the
Potato Holiday
Contest in evolving neural network controllers for the Khepera robot. The
grand prizes are three 1/2 boxes of Carlsberg, Tuborg or Ceres beers !!!
Children playing with Khepera and LEGO robots:
Here are photos of children playing
with our Khepera robot and LEGO robot demonstrations that we have shown during
different multimedia events and exhibitions.
Robot Football Competition:
The final project in the course is the
development of a robot footballer, and the participation in the Danish
Championship in Robot Football. There is a lot of informations about the Danish
Championship in Robot Football at the web site:
You can also look at the English version
We will follow the rules of the Second
Autonomous Robot Football Tournament. You might find it useful to read also
about the First Autonomous Robot Football Tournament that was held in Brighton
in July 1997 during the European Conference on Artificial Life. First Autonomous Robot Football
Tournament, and Second
Autonomous Robot Football Tournament.
Here are photos
of a Khepera robot footballer.
IR, Ambient Light, and K213 Camera Sensors:
Thomas has written a small
program that allows you to continuously monitor the different sensor inputs
(including input from the K213 vision module) and steer the robot around. It
also includes some simple behaviours that you might allow the robot to use. You
can download
khep_sensors.tar.gz. It runs in serial communication 38400 baud mode (mode
3).
Khepera K213 Vision Module:
It is a bit tricky to figure out the actual
functionality of the light intensity sensor and the 64*1 linear array sensor of
the K213 vision module, so here is an explanation of K213 vision
module.
Journals:
Conferences:
go back to Henrik Hautop
Lund's Homepage.