| |
Autonomous Object Learning using Robotic Manipulation
-- Paper by Wai Ho Li and Lindsay Kleeman
Video Walkthroughs
The following videos goes through the autonomous object learning processes, including visualizations of experiment data such as object tracking and segmentation results. These videos are a part of the first author's thesis documentation and includes a verbal narration of the autonomous learning process.
Video Walkthrough 1: White Bottle
Video Walkthrough 2: Blue Cup
Object Learning Videos and Object Recognition Results
The following table contains HDV camcorder videos of our robot performing autonomous object learning on seven test objects. The names of the test objects are the same as those used in the paper. The long pause after the nudge in the object learning videos is due to the saving of data to document the experiment. This includes the saving of several hundred tracking images collected during the robotic nudge, which takes a considerable amount of time.
The object recognition results are organized next to the corresponding object learning video. Four test images are used for each object. The object recognition results are available via the links numbered from 00 to 03. Each result image shows the test image above the matching training image returned by the recognition system. The recognized object label is printed in green. The red lines represent the SIFT descriptor matches between the test image and the training image.
Note:
The object learning videos are encoded using the MPEG-4 codec. If you are having video playback problems, try using the VLC player. It is an open source and cross platform player that can play nearly all existing video formats. Make sure to save the videos to your hard drive before playing them.
<-- See other IRRC Videos
<-- Return to IRRC Index
| |
