CS81 Developmental Robotics

• Ryan Budish and Elizabeth Holman. Who moved the cheese? Robots foraging together in a dangerous world.
• Julie Corder and Ross Messing. The combination of competitive evolution and an interactive environment.
• Hollis Easter and Frederick Heckel. Learning and evolution: A comparison.
• Feng He and Kuzman Ganchev. Associating images of simple objects with labels using image compression and an evolved neural network.
• David Kaczorowski and Andy Zuppann. Evolving cooperation: The butt lovers dilemma.
• Eric Levy. Symbol Self-Localization in Mobile Autonomous Robots Prediction Networks and Self-Organizing Maps
• Ben Mitchell and BDan Fairchild. Jack of all trades, master of none: Evolving versatility in modular connectionist robots.
• Renuka Nayak and Laura Fox. Learning and evolution in a continuous environment.
• Sven Olsen and Matt Fowles. Giving Nolfi a fair fight: A comparison of emergent modularity and Elman recurrent networks.
• Branen Salmon. Embodied evolution in a morphologically heterogeneous population of robots.
• Jeremy Stober and Jonah Gold. Evolvable morphologies for robot controllers.
• Andrew Stout and Yee Lin Tan. Toward self-organized place recognition.

Course description

This course will focus on the special topic of developmental robotics, a newly emerging paradigm of research. The goal of this research is to create intelligent robots by allowing them to go through a developmental process, rather than being directly programmed by human engineers. By endowing a robot with an appropriate initial control architecture and adaptive mechanisms, it learns through continual interactions with the world, developing self-organized mental structures. We will be studying the following sorts of questions. What should be innate in the robot? What adaptive mechanisms are needed? What motivates the robot to act? Would such a system need emotions?

The course is made-up of two components: seminar discussions and laboratory work. Each class meeting, we will spend the first half having a discussion about the assigned readings and the second half experimenting with various robot control methods.

Books & Readings

• Elman, Bates, Johnson, Karmiloff-Smith, Parisi, and Plunkett (1996). Rethinking innateness: A connectionist perspective on development, MIT Press, Cambridge, MA.
• Articles provided online or in reading packets.

Evaluation

• Class Participation & Reaction Notes: 20%

  Your reactions to each week's readings will be due by 8:00am on the morning of class. You will be posting your reactions to the online courseware package Edventure. Your username on this system will be the same as the username on your College email account. To get your password to the system, send an email from your College account to edventure@dangermouse.brynmawr.edu with a subject of password. Once you post your reaction, you will be able to read other students' reactions for that week. You should read them before class time and be ready to respond to at least one question or comment raised by someone else. Your reactions should not be summaries of the papers, instead they should be the product the reading process (e.g., questions that were raised, points that were not clear, connections to previous material you've read, etc). You are expected to be an active participant in class discussions and should come prepared for class every week.

• Labs: 20%

  The labs will introduce you to a variety of robot control methods. We will be using a system called Pyro, which stands for Python Robotics. Pyro allows you to experiment with various robots and robot simulators while only having to learn one interface. You will write the control programs in the Python programming language. For a good introduction to Python, try How to think like a computer scientist: Learning with Python.

• Midterm Project: 20%

  There will be a mid-semester paper assigned the week of February 17th and due the week of March 3rd. The goal of this midterm project is to give you a preview, on a smaller scale, of what will be expected on the final project.

• Final Project: 40%

  You will be asked to design a project related to developmental robotics. During the first half of the semester you should be thinking about possible project topics. These projects may be carried out alone or as groups of two or three students. A project proposal is due in class the week of March 24th. Each student or team will be asked to present their project to the class at the end of the semester, either during the week of April 14th or the week of April 21st. Each student or team will also turn in a research paper describing the project and the results, including a bibliography of related work. The final project paper is due on Friday, May 16 by noon.

Schedule

WEEK	SECT.2	SECT.1	TOPIC	READING	ANNOUNCEMENTS	LAB
1	1/20	1/22	Introduction	none	-	Introduction
2	1/27 Dave	1/29 Renuka	Development	Rethinking Innateness Preface & Ch. 1	Drop/add ends Friday 1/31	DirectControl
3	2/3 Matt	2/5 Julie	Connectionism	Rethinking Innateness Ch. 2 Fodor's review of the book	-	NeuralNetworks
4	2/10 Jonah	2/12 Ross	Connectionist Applications	Rethinking Innateness Ch. 3	-	SelfOrganizingMap Khepera SOM vision experiments
5	2/17 Lisa	2/19 Elizabeth	Object Permanence and Labeling Images	Week 5 papers	Midterm project assigned	Midterm Project Vision
6	2/24 Andy	2/26 Eric	Developmental robotics	Week 6 papers	-	Continue midterm project NeuralNetworksAdvanced
7	3/3 Dan	3/5 Branen	Extracting information from sensory data	Week 7 papers	Midterm project due by 3pm, Friday 3/7	Finish midterm project
-	-	-	Spring Break 3/10-3/14	-	-	Retry or extend an experiment from midterm
8	3/17 Feng	3/19 Yee Lin	Evolutionary Robotics	Week 8 papers	-	-
9	3/24 Fritz	3/26 Lisa	Interaction between learning and evolution	Week 9 papers	Last day to withdraw Friday 3/28	EvolutionaryAlgorithms
10	3/31 Jeremy	4/2 Andrew	Evolving robot brains and bodies	Week 10 papers	Project proposals due in class	-
11	4/7 Ben	4/9 Laura	Extracting information from prediction L-systems	Week 11 papers	-	-
12	4/14 Hollis	4/16 Kuzman	Embodied evolution	Week 12 papers	-	Multirobot
13	4/21 Andy	4/23 Sven	Projects	Monday Presentations: Jeremy and Jonah Feng and Kuzman	Wenesday Presentations: Laura and Renuka Julie and Ross Kuzman and Feng	Presentation guidelines Final project tips
14	4/28 Lisa	4/30 Ryan	Projects	Monday Presentations: Dan and Ben Fritz and Hollis Dave and Andy	Wednesday Presentations: Eric YeeLin and Andrew Ryan and Elizabeth Sven and Matt Branen	-
Finals	-	-	No final exam.	-	Project due by 9am, Monday 5/19	-

Week 5 Papers

• Y. Munakata, J. McClelland, M. Johnson, and R. Siegler (1997). Rethinking infant knowledge: Toward an adaptive process account of successes and failures in object permanence tasks. Psychological Review, Volume 104, Number 4. Available online.
• K. Plunkett, C. Sinha, M. Moller, and O. Strandsby (1992). Symbol grounding or emergence of symbols? Vocabulary growth in children and a connectionist net. Connection Science, Volume 4, Numbers 3 and 4. Available as a hard copy outside my office.
• D. Mareschal, K. Plunkett, and P. Harris ((1995). Developing object permanence: A connectionist model. In Proceedings of the Seventeenth Annual Conference of the Cognitive Science Society, edited by J.D Moore and J.F. Lehman, pages 170-175. Available online.

Week 6 Papers

• J. Weng, J. McClelland, A. Pentland, O. Sporns, I. Stockman, M. Sur, and E. Thelen (2001). Autonomous Mental Development by Robots and Animals. Science, Volume 291, Number 5504. Available online.
• D. Blank, D. Kumar, and L. Meeden (2002). Bringing up robot: Fundamental mechanisms for creating a self-motivated, self-organizing architecture. Presented at the conference on Simulation and Adaptive Behavior. Available online.

Week 7 Papers

• P. Cohen, T. Oates, N. Adams, and C. Beal (2001). Robot Baby 2001. Invited talk at the Twelfth International Conference on Algorithmic Learning Theory. Available online.

Week 8 Papers

• L. Meeden and D. Kumar (1998). Trends in evolutionary robotics. In Soft Computing for Intelligent Robotic Systems, edited by L.C. Jain and T. Fukuda, Physica-Verlag, New York, NY, pages 215-233. Available online.
• S. Nolfi (1997). Using emergent modularity to develop control systems for mobile robots. Adapdive Behavior, Volume 5, Number 3-4, pages 343-364. Available online.
• S. Nolfi and D. Marocco (2002). Evolving robots able to visually discriminate between objects with different sizes. International Journal of of Robotics and Automation, Volume 14, Number 4, pages 163-170. Available online.

Week 9 Papers

• S. Nolfi and D. Floreano (1999). Learning and Evolution. Autonomous Robots, Volume 7, Number 1, pages 89-113. Available online.

Week 10 Papers

• J. Pollack, H. Lipson, G. Hornby, and P. Funes (2001). Three generations of automatically designed robots, Artificial Life, Volume 7, Number 3. Available online.
• F. Dellaert and R. D. Beer (1994). Toward an Evolvable Model of Development for Autonomous Agent Synthesis. In the Proceedings of the Artificial Life IV Conference. Available online.
• K. Sims (1994). Evolving 3D mophology and behavior by competition. In the Proceedings of the Artificial Life IV Conference. Available here.

Week 11 Papers

• S. Nolfi and J. Tani (1999). Extracting regularities in space and time through a cascade of prediction networks: The case of a mobile robot navigating in a structured environment. Connection Science, Volume 11, Number 2, pages 1131-1141. Available online.
• F. Linaker and L. Niklasson. Sensory flow segmentation using a resource allocating vector quantizer. Available online.
• A short online introduction to L-systems
• G. Hornby and J. Pollack (2001). Evolving L-Systems to Generate Virtual Creatures. Computers and Graphics, Volume 25, Number 6, pages 1041-1048. Avalilable online.

Week 12 Papers

• R.A. Watson, S.G. Ficici, and J.B. Pollack (2000). Embodied evolution: Distributing an evolutionary algorithm in a population of robots. Technical report CS-00-208. Available online.