Initial word... Robogames
Initial word... Robogames
Initial word... Robogames The "CS graveyard"
Final words: project schedule … This week: example talk and a bit on Turing Machines (final hw due 5/1) Next week: project work during class (join in!) (final class meeting) Wed. 5/6: project presentations (~10 min. per project) 10-15 minute talk (example this week…) • (by email) Milestone : due 5/8/15 Your source-code to date (showing off your library) • One-page write-up of progress so far + plans still to-go • Include your project presentation slides! • (by email) Final Project : due by 5/18/15 (or earlier, if you'd like!) 2-3 page write-up of results, +s, -s, and all of your source code •
Final project FSM! still still works broken it's broken stop adding comment out features + start print statements adding print + start adding statements it works more features
State-machine limits ? Are there limits to what FSMs can do? PageRank Robotics they can't necessarily drive safely... Page Rank But are there any binary-string problems that FSMs can't solve?
State-machine limits ? Are there limits to what FSMs can do? PageRank Robotics they can't necessarily drive safely... Page Rank But are there any binary-string problems that FSMs can't solve?
State-machine limits ? 0 1 # of 0s == # of 1s rejected accepted Let's build a FSM that accepts bit strings 01100 011001 with the SAME NUMBER of 0s as 1s 01110 0110 0100 10 000 λ this last string is empty
FSMs are limited… So, let's build a better machine! ������ Turing Machine
the input 0 1 0 1 0 the tape R/W "blanks" are head everywhere else an accepting state always halts -- then basks in its success! if a transition is missing, the input FAILS! 0 ; 1 , R a Turing Machine rule: READ WRITE MOTION try it in JFLAP...
Try it! Run this TM on this input: 0 0 1 1 1 Is this input accepted or rejected? What inputs are accepted in general ? How does it work? Extra: How could you change this to accept palindromes? (a thought experiment – and ex. cr.)
Turing Machine machines !
Final words: project schedule … This week: example talk and a bit on Turing Machines (final hw due 5/1) Next week: project work during class (join in!) (final class meeting) Wed. 5/6: project presentations (~10 min. per project) 10-15 minute talk (example this week…) • (by email) Milestone : due 5/8/15 Your source-code to date (showing off your library) • One-page write-up of progress so far + plans still to-go • Include your project presentation slides! • (by email) Final Project : due by 5/18/15 (or earlier, if you'd like!) 2-3 page write-up of results, +s, -s, and all of your source code •
10-15 min. talk guideline (feel free to alter to suit!)
10-15 min. talk guideline (feel free to alter to suit!)
10-15 min. talk guideline (feel free to alter to suit!) Title + Intro "Big Picture" What's the problem ? overall context + motivation What's your What library or Problem details approach ? libraries did you (narrative > bullets) explore/use? to reducing the problem How did you adjust What worked Your testing -- to what was and what didn't? and results feasible vs. not? insights! Things you're Questions… Further progress still planning to and/or insights work on… anything else…?
"Aside Slides" blue! Interlaced notes about the slides… Suggestions and ideas… Stuff to ignore… This is a talk that summarizes a robotics and vision research project we worked on…
������������ ����������������������������� Nicole Lesperance ’11 Steve Matsumoto ’12 Kenny Lei Michael Leece ’11 Max Korbel ’13 Zach Dodds REU Harvey Mudd College – TePRA – 4/12/11
Learning distance from texture Idea: use monocular texture to create a traditional range scan Zach Dodds 4/22/2015 IST338
"Title" slide … You don't need a fancy title I'm a fan of explanatory pictures… I'm less a fan of using clip art… really? Never hurts to have a short "big-picture" summary…
Big picture: obtain distances from an image Looking at this image, we (humans) know how far away obstacles are from the camera, relatively …
"Big picture" slide What's the problem's overall goal Overall context Why it's interesting You may have a lot to say, or just a little, but include at least one such slide… I'll help you brainstorm. You do not need "alternative approaches"
Context: other approaches active lighting depth, cost, weight
��������� ����������������������������� ������������������� depth, cost, weight
��������� ����������������������������� ��������������������� depth, cost, weight
������������ ������������ to estimate: how and δ δ δ δ how much the image feature moved previous image next image ∆ ∆ ∆ ∆ to estimate: how and ����� �������� how much the camera �������� (or robot) moved texture vs. time
!���������� Could we improve (����� #�����$�%)* resolution? Could we handle more general indoor textures? "��������#�����$�%&'
+������������ Could we benefit from larger patches of image texture? �������� 3�������3������� ������--- 45��/+(������ ,-���������� �����-.�!,/0�1))2
The problem : there's lots of floor! indoor robot + webcam on netbook
The data : lots of image patches! sometimes the floor != the obstacles
The dataset : lots of patches! sometimes not
The dataset : lots of patches! not always interesting image patches…
Image patches Each 20x20 pixel square is a single "patch" hand-segmented image
Image details… Each 20x20 pixel square is a single "patch" ~ 1200 numbers each! hand-segmented image
����������6� ����������#3���7���383��������3$ (1) Train groundplane vs. (2) Classify new images using obstacle textures. groundplane/obstacle models (3) Segment new images to (4) Unwarp segmentations find traversibility hypotheses into range scans
"Problem" slides What's the computational problem? Insights into its value… Insights into its difficulty… Highlight your contributions, e.g., columns Is it labeled (how? by hand?) Can we see some of it?
Library explored… OpenCV … watched it form and grow …
Well documented! OpenCV
9���������:���,; ��� ������������������������#�������$ average average average blue green red 138 87 53 20x20 texture patches, each represented by a few values. For example, average Red, Green, and Blue (out of 255)
<�����������������7���� 20x20 texture patches, distilled into 8 values obstacle 194 191 211 3.2 25.6 4.1 ... ground 138 87 53 -1.14 8.6 1.4 ... ���� ��������������� ��������.������ ����� 2����������������
" features " <�����������������7���� texture + color filters Each 20x20 pixel square is a single "patch" ../TrainingImages/Playspacepswo13Patches/00029/randomBelow/0009.png 194.2575 191.4525 211.4775 195.0 192.0 212.0 8.6707 8.7688 7.2910 211.4775 191.4525 194.2575 212.0 192.0 195.0 7.2910 8.7688 8.6707 3113.1847 2918.6196 194.70780 -8.295e-09 -2.9999 3.4887 -0.1821 1.0586 0.7981 0.1422 1.3764 1.5171 - 3.2110 25.6053 4.0897 0.2096 3.4240 7.9828 0.6552 1.4017 1.2862 -0.0688 7.9768... ( Means, medians, stdev.,… for RGB/HSV colors and texture filters )
=������ �����7����������������� ����5)�������������� �������� �������� ����5)�������������� ������ �������� />?�������������������������
=������ �����7��� ����5)�������������� �������� �������� ����5)�������������� ������ �������� />?�@������������������ 7���������������������
��� ���� ������������������������
Recommend
More recommend