CS206Evolutionary Robotics Anatomy of an evolutionary robotics experiment: 1.Create a task environment. 2.Create the robot. 3.Create the robot’s brain,

Slides:



Advertisements
Similar presentations
Approaches, Tools, and Applications Islam A. El-Shaarawy Shoubra Faculty of Eng.
Advertisements

Introduction to Neural Networks
Chapter Thirteen Conclusion: Where We Go From Here.
Bio-Inspired Optimization. Our Journey – For the remainder of the course A brief review of classical optimization methods The basics of several stochastic.
Institute of Intelligent Power Electronics – IPE Page1 Introduction to Basics of Genetic Algorithms Docent Xiao-Zhi Gao Department of Electrical Engineering.
4-1 Management Information Systems for the Information Age Copyright 2002 The McGraw-Hill Companies, Inc. All rights reserved Chapter 4 Decision Support.
Copyright (c) John Y. Cheung, 2002 ECE Recruiting,ppt Slide 1 What is an Electrical and Computer Engineer?
The Brain is Embodied and the Body is Embedded in the Environment Jeff Krichmar Department of Cognitive Sciences University of California, Irvine.
Mental Development and Representation Building through Motivated Learning Janusz A. Starzyk, Ohio University, USA, Pawel Raif, Silesian University of Technology,
1 Chapter 4 Decision Support and Artificial Intelligence Brainpower for Your Business.
Machine Creativity. Outline BackgroundBackground –The problem and its importance. –The known algorithms and systems. Summary of the Creativity Machine.
Summer 2011 Wednesday, 8/3. Biological Approaches to Understanding the Mind Connectionism is not the only approach to understanding the mind that draws.
Outline of Presentation n What is IE/OR? n IE/OR at Umass Amherst n Department History at Umass n General Information n IE/OR Career Areas n IE/OR Related.
COMPUTER SCIENCE 10: INTRODUCTION TO COMPUTER SCIENCE Dr. Natalie Linnell with credit to Cay Horstmann and Marty Stepp.
Artificial Intelligence
Artificial Intelligence
Engineering or Mechanical Engineering?
Vedrana Vidulin Jožef Stefan Institute, Ljubljana, Slovenia
Gerhard K. Kraetzschmar The Cool Science Institute Educational Robotics A Glimpse on Robotics Tutorial Material.
Introduction to Computer and Programming CS-101 Lecture 6 By : Lecturer : Omer Salih Dawood Department of Computer Science College of Arts and Science.
Succeeding with Technology Information, Decision Support… Decision Making and Problem Solving Management Information Systems Decision Support Systems Group.
ICT Cambridge National. Content Assessment = 75% coursework ◦Includes 2 teacher assessed assignments. ◦1 Controlled Conditions assessment. ◦1 Examination.
Chapter 10. Global Village “… is the shrinking of the world society because of the ability to communicate.” Positive: The best from diverse cultures will.
EGS 1001C Introduction to Engineering Succeeding in the Classroom Professor: Dr. Miguel Alonso Jr.
Nilufa Rahim C2PRISM Fellow Sept. 12, What is Engineering? Engineering is the field of applying Science and Mathematics to develop solutions that.
Emergent Inference, or How can a program become a self-programming AGI system? Sergio Pissanetzky Self-programming Workshop AGI-11.
Advances in Robotics and How They Apply to Learning.
Biologically Inspired computing Info rm atics luis rocha 2007 biologically-inspired computing.
Copyright © 2006, The McGraw-Hill Companies, Inc. All rights reserved. Decision Support Systems Chapter 10.
Chapter 1. Introduction in Creating Brain-Like Intelligence, Sendhoff et al. Course: Robots Learning from Humans Jo, HwiYeol Biointelligence Laboratory.
Study on Genetic Network Programming (GNP) with Learning and Evolution Hirasawa laboratory, Artificial Intelligence section Information architecture field.
1 Adrian Stoica Jet Propulsion Laboratory ehw.jpl.nasa.gov Evolvable Hardware for Automated Design and Autonomous.
Mehdi Ghayoumi MSB rm 132 Ofc hr: Thur, a Machine Learning.
Computational Intelligence II Lecturer: Professor Pekka Toivanen Exercises: Nina Rogelj
Artificial Intelligence By Michelle Witcofsky And Evan Flanagan.
Research Methods Introduction to Research Methods Prof.
How Solvable Is Intelligence? A brief introduction to AI Dr. Richard Fox Department of Computer Science Northern Kentucky University.
I Robot.
University of Windsor School of Computer Science Topics in Artificial Intelligence Fall 2008 Sept 11, 2008.
Evolutionary Robotics The Italian Approach The Khepera robot (1996) Developed at EPFL Lausanne, Switzerland(!) by Francesco Mondada Diameter: 55 mm Could.
Cognitive Science: What is it, and How can I study it at RPI?
ELeaRNT: Evolutionary Learning of Rich Neural Network Topologies Authors: Slobodan Miletic 3078/2010 Nikola Jovanovic 3077/2010
Autonomous Virtual Humans Tyler Streeter. Contents Introduction Introduction Implementation Implementation –3D Graphics –Simulated Physics –Neural Networks.
M.Sc. and Ph.D. in Computational Science Department of Mathematics Faculty of Science Chulalongkorn University.
“Politehnica” University of Timisoara Course Advisor:  Lucian Prodan Evolvable Systems Web Page:   Teaching  Graduate Courses Summer.
Course Overview  What is AI?  What are the Major Challenges?  What are the Main Techniques?  Where are we failing, and why?  Step back and look at.
Robotics By: Phil FosterPhil Foster CMIS 102 Intro to Computers PowerPoint Homework Assignment 2 February 24, 2004.
A field of study that encompasses computational techniques for performing tasks that require intelligence when performed by humans. Simulation of human.
Artificial Intelligence ( AI ) Ahmed Ismail AIT 600 Spring 08.
1/24 Dr. Robert Layton Internet Commerce Security Laboratory Centre for Informatics and Applied Optimisation University of Ballarat Mathematics and Cybercrime.
Robot Intelligence Technology Lab. 10. Complex Hardware Morphologies: Walking Machines Presented by In-Won Park
Robot Intelligence Technology Lab. Evolutionary Robotics Chapter 3. How to Evolve Robots Chi-Ho Lee.
Evolving robot brains using vision Lisa Meeden Computer Science Department Swarthmore College.
Postgraduate stud. Al-Ahnomi Montaser Don State Technical University Department “Computer-aided design" Theme:- "development and research of intelligent.
Overview of Artificial Intelligence (1) Artificial intelligence (AI) Computers with the ability to mimic or duplicate the functions of the human brain.
Classification of models
MYPF 2.1 Finding the Right Career Fit 2.2 Finding Career Opportunities
Computer Science at UNCW
Chapter 11: Artificial Intelligence
CSIS 104 –Intro. To Computer Science
Done Done Course Overview What is AI? What are the Major Challenges?
Software Usability and Design
Artificial Intelligence introduction(2)
Science Fair Categories
Artificial Intelligence (Lecture 1)
Enabling ML Based Research
Principles of Computing – UFCFA3-30-1
CHAPTER I. of EVOLUTIONARY ROBOTICS Stefano Nolfi and Dario Floreano
Artificial Intelligence and Future of Education
Introduction to Artificial Intelligence Instructor: Dr. Eduardo Urbina
Presentation transcript:

CS206Evolutionary Robotics Anatomy of an evolutionary robotics experiment: 1.Create a task environment. 2.Create the robot. 3.Create the robot’s brain, or Artificial Neural Network (ANN). 4.Use an evolutionary algorithm to optimize the ANN so that the robot performs the desired task in its environment.

Who I am: 1993 – 1997: Undergraduate in Computer Science from McMaster University, Hamilton, Canada 1997 – 1998:Software Engineer, Computing Devices Canada, Calgary, Canada 1998 – 1999:MSc in Evolutionary and Adaptive Systems from Sussex University, Brighton, England 1999 – 2003:PhD in Mathematical and Physical Sciences from University of Zurich, Switzerland 2003 – 2006:Postdoctoral work at Cornell University, Ithaca, NY Present:Associate Professor, UVM CS206Evolutionary Robotics

…and you are? Freshman/junior/senior/grad, what are you majoring in? What are your hopes for this course? What are your hopes for this degree? What are your career goals? CS206Evolutionary Robotics

What I expect from you: Feedback Common sense Regular, not necessarily perfect attendance Hard work Memorization of key concepts (job interviews are closed-book) Creativity Self - learning A positive attitude when working with me, the TA, or fellow students. CS206Evolutionary Robotics

What you _cannot_ expect from me: Help with buggy code (eg. “Why is my program crashing?”) Help with learning a programming language, installing software, etc. CS206Evolutionary Robotics

What you _can_ expect from me: Help with _specific_ programming questions… (eg. “Why do I need to deallocate memory?”) … but only after you’ve consulted online resources and your peers. (Google “C++ tutorial”) Help with conceptual issues (“Can you go over the genetic algorithm again?”) Clarification about the assignments / midterm / project An emphasis on concepts, rather than specific tools, because tools change, but concepts change more slowly. CS206Evolutionary Robotics

CS206Evolutionary Robotics

Assignment 1: Create an Optimizer. Assignment 2: Create a Neural Network. Assignment 3: Evolve Neural Networks. Assignment 4: Place objects in the physics engine. Assignment 5: Add joints to the physics engine. Assignment 6: Add motors to the physics engine. Assignment 7: Add sensors to the physics engine. Assignment 8: Add the neural network to the physics engine. Assignment 9: Evolve the neural network in the physics engine. Assignment 10: Evolve behaviors in the physics engine. Weeks 9 to 14: Use the system to conduct an evolutionary robotics experiment. Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 CS206Evolutionary Robotics

Weeks 9 to 14: Use the system to conduct an evolutionary robotics experiment. 1.Given the same optimizer, neural network and eight hours of behavior optimization on the same computer, does a quadrupedal robot evolve to walk further or not as far as a hexapedal robot, or is there no significant difference? 2. Create five fitness functions that not only select for locomotion on the quadrupedal robot, but each also selects for a particular gait: walking, trotting, canter, galloping and pronking. 3. Create a fitness function that rewards NNs for locomotion, but penalizes them for requiring a lot of energy to realize the gait. This is difficult, as there are two solutions that are not desirable: evolution finds fast but inefficient gaits, or ‘gaits’ in which the robot does not move, and therefore does not consume energy. 4. Equip the robot with a simulated laser range finder, which tells the robot about objects in its environment. Evolve a robot that walks toward round objects, but walks away from rectangular objects. CS206Evolutionary Robotics

Why Robots? CS206Evolutionary Robotics

Why Robots? It is one of the few ‘open frontiers’ in science: There is no guiding theory about intelligence Guiding theory in biology: evolution Guiding theory in physics: Newtonian and quantum mechanics Guiding theory in chemisty: … … There are several Nobel Prizes waiting to be won in this field. It is one of the most interdisciplinary areas of study: Evolutionary biology Neuroscience Cognitive science Psychology Biomechanics Physics Computer Science Mechanical engineering, electrical engineering, bio-engineering Chemistry Mathematics CS206Evolutionary Robotics

Large-scale outdoor infrastructure: Wind farms, Solar farms, Wave farms, Tide farms, … Why Robots Outside? CS206Evolutionary Robotics

Why Evolutionary Robotics? 1.Creating a controller for a robot is non-intuitive: requires automation t1 t2 t3 t4 t5 t6 t7 t8 … m … m … m … m … … CS206Evolutionary Robotics

Why Evolutionary Robotics? 1.Creating a controller for a robot is non-intuitive: requires automation 2.Learning algorithms only optimize controllers; evolution can optimize the whole robot. CS206Evolutionary Robotics

Why Evolutionary Robotics? 1.Creating a controller for a robot is non-intuitive: requires automation 2.Learning algorithms only optimize controllers; evolution can optimize the whole robot. 3. Biological evolution produced adaptive agents of unparalleled complexity with no supervision. CS206Evolutionary Robotics

Why Evolutionary Robotics? 1.Creating a controller for a robot is non-intuitive: requires automation 2.Learning algorithms only optimize controllers; evolution can optimize the whole robot. 3. Biological evolution produced adaptive agents of unparalleled complexity with no supervision. 4. Most learning algorithms require detailed supervision: t1 t2 t3 t4 t5 t6 t7 t8 … m … m … m … m … … Should have been ‘+1’ CS206Evolutionary Robotics