2. Syllabus Theory Locomotion Kinematics Perception Localization Planning and Navigation Practice Simulation exercises Debates Participation in online competition Youtube
3. Links and Places Class wiki http://correll.cs.colorado.edu/pmwiki Code How-Tos and exercises CU Learn http://culearn.colorado.edu Submission of assignments Exercises CSEL 128 http://sac.colorado.edu Contact ECOT 733 (my office) Phone: (303) 492-2233 nikolaus.correll@colorado.edu Try this out asap! Try this out asap!
4. Textbook Introduction to Autonomous Mobile Robots, Roland Siegwart and Illah R. Nourbakhsh, MIT Press Available from CU Bookstore next week Weekly reading assignments Lecture is complementing the book
5. Activities and Grading Final examen Deliverables Weekly reading assignments Midterm Debate Class participation Must attend to lectures and seminars 40% 20% 20% 10% 10%
6. What will you learn? What are robots, what is the science and technology behind building robots and programming them? Why is robotics hard? Hands-on experience programming of driving, walking and crawling robots
7. What will not be covered? Because we will be working in simulation, we will not cover Components you build are less likely to work than those that you buy Components you bought are much harder to debug than those you built Software-engineering Take: Advanced robotics
8. Not be covered C / C++ / Java Computer Vision -> CSCI 5722
9. Why robots? Robots will work more efficient and safer than humans From repetitive tasks to true autonomy Improve lifestyle and education Potentially as disruptive as the wheel, printing press, steam engine, internet
16. Early robots George Devol, *1912 ~1940 ~1950 Are these robots or automatons? Which are more robots than others and why? “Unimate”, shipped to GM in 1961
17. What is a robot? Sensing perception Computation making sense of your senses Actuation and Mechanism moving and manipulation Communication communicating with people, objects and other robots
18. Who is doing robotics? Social Sciences Biology Chemistry Aerospace Engineering Physics
22. Big Dog Sensing Legs: Joint positions and force, ground contact Body: Gyroscope, Stereovision Computation/Control High-frequency closed-loop control Actuation/Mechanism 4 x 4 DOF hydraulic legs Communication Remote Control
24. Justin Sensing Arms/torso/fingers: position, force/torque Head: vision, hearing Computation Compliant control of fingers and arm Object recognition and language processing Motion planning for collision-free paths Actuation 2 x 6 DOF Kuka arm, torso, legs, 2 x 4 x 3 DOF hand Communication Aural and speech
26. Kiva Sensing Bar codes on the floor (localization), odometry Computation Centralized, grid-based algorithm Actuation Moving on the grid, lift cupboards Communication Many-to-one (centralized), user to server
27. This course Theory How do sensors and actuators work How can we describe and control a robot’s motion Algorithms for localization and navigation Practice Programming Robots in Webots Designing a robot soccer player Debating about robot technology in class
28. Webots Realistic, physics-based simulation Simulates robot dynamics, sensor noise Write controllers in Java or C Goal: write a competitive robot for ratslife
29. RatsLife Competition Upload your controller to www.ratslife.org Controller is evaluated nightly against other players worldwide Download a video of the competition the next day Upload your controller into a real robot setup
30. Ratslife You will design a controller in teams of 2 Skills Perception Navigation Planning You will implement this controller and evaluate it against your peers
31. Summary A robot becomes autonomous due to a combination of Actuation Sensing Computation Communication Robotics is an interdisciplinary effort Computer science research makes robots intelligent
32. Next Meetings / Assignments Wednesday, 4.45pm, ECCS 128 First steps in Webots Reading Webots User Manual (wiki) Questions on Chapter 2, pages 13 – 32 (culearn), due Monday, September 13, 5pm.
3: algorithms, mechanisms, sensors, signal processing, focus on multi-robot systems
2 min: Intelligence is a function of ALL things coming together: sensors, algorithms, mechanisms and communication
2 min – The Athlete robot is no doubt a robot. Focus, however, is on the mechanical design. The robot has 6 legs that can have the robot move, act as pinchers and pick up stuff, or manipulate the environments with add-on tools. There is little sensing, little computation, and little autonomy on this robot. In fact, nobody even bothered thinking about how to coordinate all the different joints (6 x6 ) to actually perform all the actions that the robot theoretically could do.
4 min. Explain Sensing actuation computation and communication. Again: focus is on a revolutionary mechanical design. My labmate at MIT MarsetteVona – a computer scientist like you – built the interface for controlling the robot. He also developed computational tools that allow to calculate all the joint positions of a robot as a function of a virtual articulation. For pinching, e.g., you can think about calculating all your joint positions as a function of contact points with a virtual object. Show video.
6 min – Big Dog, what it can do, why it was build. Interesting mechanical design (four legs with 4 DOF each). Focus here is on control: how to create a gait to move the robot forwards, and – more difficult – how to react so that the robot remains stable
2 min : explain where the meat is in the big dog
5 min : justin is putting it all together, massive sensing, state-of-the art control, mechanism, computation and communication
