1 / 22

Sensor Fusion Using Dempster-Shafer Theory

Sensor Fusion Using Dempster-Shafer Theory. Huadong Wu, Mel Siegel The Robotics Institute, Carnegie Mellon University Rainer Stiefelhagen Interactive System Labs, University of Karlsruhe Jie Yang Interactive System Labs, Carnegie Mellon University. agenda.

clattimore
Download Presentation

Sensor Fusion Using Dempster-Shafer Theory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Sensor Fusion Using Dempster-Shafer Theory Huadong Wu, Mel Siegel The Robotics Institute, Carnegie Mellon University Rainer Stiefelhagen Interactive System Labs, University of Karlsruhe Jie Yang Interactive System Labs, Carnegie Mellon University IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  2. agenda • describe sensor fusion requirement for context-aware computing applications • describe application case study: meeting participants’ focus-of-attention analysis • review alternative sensor fusion approaches • introduce Dempster-Shafer theory of evidence • develop “weighted D-S evidence combination” • demonstrate its effectiveness in the case study • consider extensions of scale and scope IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  3. outline • context aware computing • sensor fusion • system architecture • case: focus-of-attention sensor fusion • Dempster-Shafer approach • implementation details • experiments & analysis • conclusions and future work IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  4. context aware computing • best algorithm for human-computer interaction tasks depends on context • context can be difficult to discern • multiple sensors give complementary (and sometime contradictory) clues • sensor fusion techniques needed • (but best algorithm for sensor fusion tasks may depend on context!) IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  5. sensor fusion • how to combine outputs of multiple sensor perspectives on an observable? • modalities may be “complementary”,“competitive”, or “cooperative” • technologies may demand registration • variety of historical approaches, e.g.: • statistical (error and confidence measures) • voting • Bayesian (probability inference) • neural network, fuzzy logic, etc IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  6. generalizable SF architecture • for context-aware computing applications • cartoon (next) illustrates typical configuration • low level sensor fusion done locally • sensor clusters deliver local perceptions • implemented via LAN communication and shared global database technologies IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  7. user context database sensor Internet smart sensor node sensor gateway smart sensor node sensors appliance applications sensor fusion embedded OS Intranet appliance embedded OS sensor applications lower-level sensor fusion context server higher-level sensor fusion appliance smart sensor node database sever embedded OS site context database sensors system architecture to support sensor fusion for context-aware computing IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  8. dynamic database • example: user identification and tracking • tables (next) list basic information about environment (room) and parameters, e.g., • temperature, noise, lighting, available devices, number of people, segmentation of area, etc • detail: analyze focus-of-attention of each user in a meeting at a small conference table IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  9. context information architecture:dynamic context information database IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  10. system architecture • create a “sensor fusion mediator” for each context • mediator collects and processes sensor data • in this test, sensor fusion implemented using Dempster-Shafer theory of evidence • need to develop differential weighting scheme for sensors in which we have differing confidence • allows combination of individual sensors’ observation with specified confidence level • admits measures of ignorance as well as knowledge IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  11. Other AI algorithms Other AI algorithms Interpreter Interpreter application application context data context data Resource Registry Resource Registry AI algorithms Dempster-Shafer rule Dynamic Context Database Discoverer Widget Widget Widget Widget sensor sensor sensor sensor SF mediator SF mediator user - mobile site context site context Aggregator Aggregator computer server database server system configuration IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  12. focus of attention • recorded meeting activities of four people around a conference table • “whose focus-of-attention is on whom” ground-truth found manually at 10 Hz • microphones detect who is talking • video imagery analyzed for each talker/listener pair IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  13. Perspective View Camera View focus of attention analysis: equipment Panoramic View IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  14. video image analysis • neural network estimates head poses • focus of attention estimate based on head pose probability distribution analysis • audio reports speaker, assumed to be focus of other participants’ attention • situation is not easy to analyze due to, e.g., dependence of behavior on discussion topic • suggests we need more general fusion approach than provided by Bayesian IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  15. focus of attention:estimation from video and audio sensors IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  16. comparison of sensor fusion alternatives #1. complementary #3. cooperative Parametric template, Figures of merit, Syntactic pattern recognition … … Logical template AI rule-based reasoning, Heuristic inference Neural network … … #2. competitive IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  17. add Dempster-Shafer approach • generalization of Bayesian approach • “theory of evidence” • implement via quantitative definitions of “belief” and “plausibility” IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  18. sensor fusion with Dempster-Shafer theory of evidence algorithm • Frame of discernment Θ: { {L}, {O}, {R}, {L | O}, {O | R}, {L | R}, {L | O | R} } • Updated Belief: IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  19. we add: weighted Dempster-Shafer evidence combination rule IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  20. weight == prior probability IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  21. person validframes audiocorrect videocorrect linearsum DScorrect weightedDS Experiment Set2 #0 1229 55.6% 70.5% 70.1% 70.0% 71.4% #1 1075 61.5% 66.2% 69.8% 70.0% 69.4% #2 1098 66.1% 78.3% 80.2% 80.8% 80.2% #3 991 68.8% 60.3% 65.6% 66.6% 70.0% Experiment Set5 #0 768 73.8% 74.4% 76.8% 77.0% 77.0% #1 956 67.6% 68.5% 72.0% 72.3% 72.1% #2 1006 73.2% 83.0% 84.1% 84.2% 83.9% #3 929 53.3% 67.9% 75.7% 76.9% 73.2% Experiment Set6 #0 799 59.1% 70.6% 71.2% 71.5% 71.0% #1 751 63.3% 84.6% 85.5% 85.8% 85.2% #2 827 57.4% 82.2% 83.3% 84.3% 83.4% #3 851 60.8% 80.6% 81.9% 82.3% 81.7% Experiment Aufnahme2 #0 653 73.2% 84.2% 85.0% 85.0% 84.2% #1 653 57.3% 53.5% 54.2% 54.2% 54.5% #2 681 72.7% 65.6% 69.5% 69.3% 70.3% #6 435 85.3% 75.6% 78.2% 78.4% 79.8% summary 13702 64.6% 72.8% 75.8% 75.4% 75.4% Sensor Fusion Results of Focus-Of-Attention Experiments IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

  22. conclusions and future research • Dempster-Shafer evidence combination rule and weighted Dempster-Shafer evidence combination rule are generalized forms of the classic Bayesian inference method • Dempster-Shafer method, especially the weighted Dempster-Shafer method, is suitable for sensor fusion tasks in context-sensing architectures with highly dynamic sensor configurations • we expect the weighted Dempster-Shafer evidence combination rule will outperform linear summation and other sensor fusion methods in planned research that will involve additional sensors IMTC-2002-1076 Dempster-Shafer mws@cmu.edu

More Related