1. CPS 001 1 Topics since last test l Graphics l Software design  Recursion  Arrays  Copyright issues l Computer systems  Hardware  Architecture  Operating Systems  Security l Computer Science Theory  Performance of algorithms  Complexity  Computability l Debate Topics

2. CPS 001 2 The exam l Tuesday, May 2, 2pm-5pm in B101 LSRC l Open book/open note l ~40% multiple choice/short answer l Cumulative l By Friday, April 28 at 5pm:  All grades up (except final project)  All solutions out  Grade problems: Submit throught Eclipse assignment name issues Final grades up Friday, May 5

3. CPS 001 3 Help Sessions 1. Sun 3-5pm 2. M 5:30-6:30 In D106 LSRC (look for email reminder)

4. CPS 001 4 Essential concepts There is beauty at all levels of sophistication and all levels of abstraction. - David A. Blackwell If life were really fair, algebra would actually come in handy - Amstel Light commercial

5. CPS 001 5 On programming and deadlines Observe that for the programmer, as the chef, the urgency of the patron may govern the scheduled completion of task, but it cannot govern the actual completion. An omelet, promised in two minutes, may appear to be progressing nicely. But when it has not set in two minutes, the customer has two choices -- wait or eat it raw. Software customers have the same choices.. - Fred Brooks We don’t have time to stop for gas -- we’re already late. - Old software project planning proverb via Mike Cleron I love deadlines. I like the whooshing sound they make as they fly by. - Douglas Adams

6. CPS 001 6 Why is programming fun? What delights may its practitioner expect as a reward? First is the sheer joy of making things Second is the pleasure of making things that are useful Third is the fascination of fashioning complex puzzle- like objects of interlocking moving parts Fourth is the joy of always learning Finally, there is the delight of working in such a tractable medium. The programmer, like the poet, works only slightly removed from pure thought-stuff. Fred Brooks

7. CPS 001 7 On education The college you attend does not determine the scope and possibility of your life’s achievements. It will have some influence, no doubt. What is more important is the encouragement that we, as parents and friends, offer these prospective students as they explore their own educational trail. In the end, the experiences they encounter and the depth of character they build along the way will mean far more than the name of the institution on their diploma. - John Hennesy Education is not filling a bucket but lighting a fire. - William Yeats

8. CPS 001 8 On education An education isn’t how much you have committed to memory, or even how much you know. It’s being able to differentiate between what you know and what you don’t. - Anatole France The best way to have a good idea is to have lots of ideas. - Linus Pauling If there is no struggle, there is no progress - Frederick Douglass The ability to quote is a serviceable substitute for wit. - W. Somerset Maugham

9. CPS 001 9 Who are these people? Why are they important?

10. CPS 001 10 Who are these people? Why are they important?

11. CPS 001 11 Who are these people? Why are they important?

12. CPS 001 12 Laws governing computer science l Moore’s Law (1965)  The number of transistors per area on a chip double every 18 months  Density of transistors => more functionality and speed l How about multiple computers? l Amdahl’s Law (1967)  Given: fraction ( s ) of work to be done is serial (i.e. isn’t parallelizable)  Maximum speedup with infinite number of processors is 1/s

13. CPS 001 13 What are computers for? l Simulation l Communication among people  Storage = communication across time l Control  Get physical  Get real (time)  Get mobile

14. CPS 001 14 Application l Simulation  Models of the real world (e.g. planets, cities, molecules) l Communication among people  Information at your fingertips  Telepresence  Home l Control  Robots  Software agents

15. CPS 001 15 What’s next l CompSci 4  Robots  Video games  Java l CompSci 6  Assumes knowledge of loops & arrays l Seminars  Animation and virtual worlds  History of Communication l Interdisciplinary minor  Computational Biology & informatics  Computational Economics

16. CPS 001 16

17. CPS 001 17 NYTimes in 1984

18. CPS 001 18

19. CPS 001 19

20. CPS 001 20 What do I do? Robotics & AI l Making systems that act rationally l Interesting problems  Robocup  Autonomous vehicle control

21. CPS 001 21 Robocup l The Goal: By the year 2050, develop a team of fully autonomous humanoid robots that can win against the human world soccer champion team

22. CPS 001 22 Robocup Rescue l Goal: When disaster happens, minimize risk to search and rescue personnel, while increasing victim survival rates, by fielding collaborative teams of robots that can:  Autonomously negotiate compromised and collapse structures  Find victims and ascertain their conditions  Produce practical maps of their locations  Identify hazards  Deploy sensors (acoustic, thermal, hazmat, seismic,...)  Provide structural shoring l... allowing human rescuers to quickly locate and extract victims

23. CPS 001 23 Outreach l How can we use robots to inspire middle school students? l What about the Digital Divide? l RoboCupJunior: International, national, and regional competitions for elementary, middle, and high school students

24. CPS 001 24 Autonomous Vehicle Control: The World Noisy, unpredictable and hostile Delayed feedback from actions Partially Observable  Significant challenge for AI

25. CPS 001 25 Autonomous Vehicle Control: Approaches l Vehicle control  Significant progress on low-level sensing and control  [Dickmans and Zapp, 1987; Godbole and Lygeros, 1993; Pomerleau, 1993; Malik et al., 1997] l High-level reasoning  SAPIENT system at CMU  RL methods  The Bayesian Automated Taxi Project Goal: Develop and test control architecture for learning control of autonomous vehicles Domain: Simulator –Much safer!

26. CPS 001 26 Autonomous Vehicle Control: Learning to Drive l No single optimal trajectory or path  Not easily amenable to supervised learning or regulatory control methods  Developed an explicit policy representation for control which performed robustly in a number of driving scenarios  Somewhat fragile and not easily adaptable l Reinforcement learning (RL)  Successively improves and adapts control strategies through trial-and-error interactions with a dynamic environment l RL techniques have shown some promise in solving complex control problems  TD-Gammon [Tesauro, 1992], Inverted helicopter control [Ng, 2004], Dialogue management, Intelligent tutoring systems  Need to scale and extend for continuous control domains

27. CPS 001 27 Reinforcement Learning in MDPs l Markov Decision Problems / Optimal control  Theoretical framework for controllers to maximize some external performance criteria l Definitions:  State - A particular situation in the world as viewed by agent  Policy -What to do in every state  Model -What follows what  Reward -What is good Agent Environment Action u t State s t Reward r t

28. CPS 001 28 What is SLAM? l Simultaneous Localization and Mapping l Localization  Finding one’s place within a map  Typically assumes a map  Uses only built-in sensors (no GPS!)  (Relatively) easy with 100% accurate map l Mapping  Building a representation of the world  (Relatively) easy with 100% accurate localization

29. CPS 001 29 Example Applications l Planetary exploration l Search and Rescue l Hostage/terrorist situations l De-mining (land/sea) l Blueprint correction l Need robot’s eye view of the world

30. CPS 001 30 Example Robot Markov the robot, generously donated by SAIC

31. CPS 001 31 Example Map

32. CPS 001 32 Why is SLAM hard: Odometry Actual trajectory Odometry

33. Why is SLAM Hard: Ambiguity Start End Same position

34. CPS 001 34 DARPA Grand Challenge l Follow a route l Avoid obstacles l Win $2 million!