Change search
ReferencesLink to record
Permanent link

Direct link
Adaptive Supervision Online Learning for Vision Based Autonomous Systems
Linköping University, Department of Electrical Engineering, Computer Vision. Linköping University, Faculty of Science & Engineering.
2016 (English)Doctoral thesis, monograph (Other academic)
Abstract [en]

Driver assistance systems in modern cars now show clear steps towards autonomous driving and improvements are presented in a steady pace. The total number of sensors has also decreased from the vehicles of the initial DARPA challenge, more resembling a pile of sensors with a car underneath. Still, anyone driving a tele-operated toy using a video link is a demonstration that a single camera provides enough information about the surronding world.  

Most lane assist systems are developed for highway use and depend on visible lane markers. However, lane markers may not be visible due to snow or wear, and there are roads without lane markers. With a slightly different approach, autonomous road following can be obtained on almost any kind of road. Using realtime online machine learning, a human driver can demonstrate driving on a road type unknown to the system and after some training, the system can seamlessly take over. The demonstrator system presented in this work has shown capability of learning to follow different types of roads as well as learning to follow a person. The system is based solely on vision, mapping camera images directly to control signals.  

Such systems need the ability to handle multiple-hypothesis outputs as there may be several plausible options in similar situations. If there is an obstacle in the middle of the road, the obstacle can be avoided by going on either side. However the average action, going straight ahead, is not a viable option. Similarly, at an intersection, the system should follow one road, not the average of all roads.  

To this end, an online machine learning framework is presented where inputs and outputs are represented using the channel representation. The learning system is structurally simple and computationally light, based on neuropsychological ideas presented by Donald Hebb over 60 years ago. Nonetheless the system has shown a cabability to learn advanced tasks. Furthermore, the structure of the system permits a statistical interpretation where a non-parametric representation of the joint distribution of input and output is generated. Prediction generates the conditional distribution of the output, given the input.  

The statistical interpretation motivates the introduction of priors. In cases with multiple options, such as at intersections, a prior can select one mode in the multimodal distribution of possible actions. In addition to the ability to learn from demonstration, a possibility for immediate reinforcement feedback is presented. This allows for a system where the teacher can choose the most appropriate way of training the system, at any time and at her own discretion.  

The theoretical contributions include a deeper analysis of the channel representation. A geometrical analysis illustrates the cause of decoding bias commonly present in neurologically inspired representations, and measures to counteract it. Confidence values are analyzed and interpreted as evidence and coherence. Further, the use of the truncated cosine basis function is motivated.  

Finally, a selection of applications is presented, such as autonomous road following by online learning and head pose estimation. A method founded on the same basic principles is used for visual tracking, where the probabilistic representation of target pixel values allows for changes in target appearance.

Place, publisher, year, edition, pages
Linköping: Linköping University Electronic Press, 2016. , 176 p.
Linköping Studies in Science and Technology. Dissertations, ISSN 0345-7524 ; 1749
National Category
Computer Vision and Robotics (Autonomous Systems)
URN: urn:nbn:se:liu:diva-125916DOI: 10.3384/diss.diva-125916ISBN: 978-91-7685-815-8 (Print)OAI: diva2:916645
Public defence
2016-05-20, Visionen, B-building, Campus Valla, Linköping, 09:00 (English)
EU, FP7, Seventh Framework ProgrammeSwedish Research Council
Available from: 2016-04-19 Created: 2016-03-08 Last updated: 2016-05-16Bibliographically approved

Open Access in DiVA

fulltext(5398 kB)318 downloads
File information
File name FULLTEXT02.pdfFile size 5398 kBChecksum SHA-512
Type fulltextMimetype application/pdf
omslag(361 kB)30 downloads
File information
File name COVER01.pdfFile size 361 kBChecksum SHA-512
Type coverMimetype application/pdf
Supplementary files with videos(62 kB)13 downloads
File information
File name ATTACHMENT03.pdfFile size 62 kBChecksum SHA-512
Type attachmentMimetype application/pdf
Channel vector curves, Four channels, 3D space(5573 kB)44 downloads
File information
File name MOVIE01.mp4File size 5573 kBChecksum SHA-512
Type movieMimetype video/mp4
Channel vector curves, Five channels, 4D space(3862 kB)16 downloads
File information
File name MOVIE02.mp4File size 3862 kBChecksum SHA-512
Type movieMimetype video/mp4
Channel vector curves, Seven channels, 6D space(6091 kB)13 downloads
File information
File name MOVIE03.mp4File size 6091 kBChecksum SHA-512
Type movieMimetype video/mp4
Cone, Three channels, 3D space(11233 kB)11 downloads
File information
File name MOVIE04.mp4File size 11233 kBChecksum SHA-512
Type movieMimetype video/mp4
Cone, Four channels, 4D space(21066 kB)16 downloads
File information
File name MOVIE05.mp4File size 21066 kBChecksum SHA-512
Type movieMimetype video/mp4
Cone, Seven channels, 7D space(37349 kB)12 downloads
File information
File name MOVIE06.mp4File size 37349 kBChecksum SHA-512
Type movieMimetype video/mp4
Associative learning illustration(2842 kB)15 downloads
File information
File name MOVIE07.mp4File size 2842 kBChecksum SHA-512
Type movieMimetype video/mp4
Decoding of five pixels in a sequence(26451 kB)12 downloads
File information
File name MOVIE17.mp4File size 26451 kBChecksum SHA-512
Type movieMimetype video/mp4
Sequence with translating cameraman image(742 kB)8 downloads
File information
File name MOVIE14.mp4File size 742 kBChecksum SHA-512
Type movieMimetype video/mp4
Video from UAV(8388 kB)13 downloads
File information
File name MOVIE15.mp4File size 8388 kBChecksum SHA-512
Type movieMimetype video/mp4
Original video from the UAV(16423 kB)13 downloads
File information
File name MOVIE16.mp4File size 16423 kBChecksum SHA-512
Type movieMimetype video/mp4
Autonomous Road Following Application, Use case demo(270078 kB)27 downloads
File information
File name MOVIE12.mp4File size 270078 kBChecksum SHA-512
Type movieMimetype video/mp4
Autonomous Road Following Application, Demonstrator system(159031 kB)34 downloads
File information
File name MOVIE13.mp4File size 159031 kBChecksum SHA-512
Type movieMimetype video/mp4

Other links

Publisher's full text

Search in DiVA

By author/editor
Öfjäll, Kristoffer
By organisation
Computer VisionFaculty of Science & Engineering
Computer Vision and Robotics (Autonomous Systems)

Search outside of DiVA

GoogleGoogle Scholar
Total: 324 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Altmetric score

Total: 4202 hits
ReferencesLink to record
Permanent link

Direct link