LectureBlog

Home Lecture Blog Resources Project Discussion Group Assignments

Meeting 20: Thu Apr 4

Meeting 19: Tue Apr 3

Meeting 18: Thu Mar 29

  • PS9 is updated
  • work in lab

Meeting 17: Tue Mar 27

  • recap on Bilibot work
  • how to get Kinect to work
  • PS9
  • discussion on beam-finding model (see photo)

Meeting 16: Thu Mar 22

  • work on PS8 in lab

Meeting 15: Tue Mar 20

  • work on PS8 in lab

Meeting 14: Thu Mar 8

  • midterm

Meeting 13: Tue Mar 6

Meeting 12: Thu Mar 1

  • snow day; no class meeting.

Meeting 11: Tue Feb 28

  • midterm is Thu Mar 8
  • PS7 is due Mar 8
  • IGVC?
  • histogram vs. particle vs. 1-D vs 2-D localization
  • 2-D histogram localization
(:youtube 9a42_zEeeA0:)
  • live coding of PS7

Meeting 10: Thu Feb 22

Meeting 9: Tue Feb 21

  • issues in planar (3-DOF) localization
  • ray tracing and correspondence of laser readings to map
  • demo of PS7 code
  • Echo360

Meeting 8: Thu Feb 16

  • how the sensing-update step is actually Bayes rule
  • next three weeks: 1d particle filter implementation (PS5-6), 3d PF implementation in sim, 3d PF on Bilibots.
  • demos of 1d grid localization solutions
  • introduction to particle filtering
(:youtube H0G1yslM5rc:)
(:youtube qQQYkvS5CzU:)
(:youtube lwg_KI3UewY:)
(:youtube eWsEyJCVXoo:)

Meeting 7: Tue Feb 14

  • demonstration of localization algorithm:
(:youtube u293629ZwIo:)

Meeting 6: Thu Feb 9

  • demonstration of PS4 solutions
  • introduction to PS5?
  • Echo360

Meeting 5: Tue Feb 7

  • robot localization Bayes network including map information
  • taxonomy of localization
  • localization methods: Kalman filters, grid/histograms, Monte Carlo/particle filters
  • white board notes (pdf)
  • Echo360

Meeting 4: Thu Feb 2

  • Introduction to writing a real program in ROS, including receiving laser scan data and controlling robot speed and rotation.
  • Discussion of PS4.
  • Echo360

Meeting 3: Tue Jan 31

  • white board notes (pdf): 2D visual explanation of Bayes Rule; from predictions to beliefs; various belief representations; 3 eqn-in-3-unknown solution to Markov process stationary state.
  • Introduction to ROS: “ROS is an open-source, meta-operating system for your robot. It provides the services you would expect from an operating system, including hardware abstraction, low-level device control, implementation of commonly-used functionality, message-passing between processes, and package management. It also provides tools and libraries for obtaining, building, writing, and running code across multiple computers.” (from http://www.ros.org/wiki/ROS)

Meeting 2: Thu Jan 26

  • Derivation of Bayes rule P(A^B) = P(B|A)P(A) = P(A|B)P(B)
  • Solving basic A->B network problem: observation of B means P(A) falls from 0.5 to 0.2
  • Solving robot seeing door or wall: prior of 0.5 goes up to ~0.9 with just one sensor reading
  • HMMs have hidden state that you can infer from their “emissions” using Bayes rule
  • At a fundamental level, the assumption of probabilistic robotics is that you can't ever know the state of the world directly—it is a hidden variable and you can only infer it from sensor readings
  • Discussed Markov assumption implicit in equations 2.31 and 2.32 of text
  • Discussed basic loop of (a) control output u influencing state x, (b) new state being probabilistically sensed as z, and then (c) new control u plus just-sensed z as influencing new state x (Figure 2.2 of text).

Meeting 1: Tue Jan 24

  • Course overview
  • Bilibot and IGVC
  • Introduction to probabilistic methods
(:youtube 8mi8z-EnYq8:) (:youtube H0G1yslM5rc:)