![Page 1: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/1.jpg)
Online Construction of Analytical PredictionModels for Physical Environments:
Application toTraffic Scene Modeling
Anurag Umbarkar, Shreyas K Rajagopal and Alex Doboli
Department of Electrical and Computer Engineering
State University of New York, Stony Brook, NY 11794
Email: [email protected]
![Page 2: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/2.jpg)
Presentation Outline
• Introduction
• Main Challenges
• Problem Description
• Modeling Methodology
• Experimental Results
• Conclusions
![Page 3: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/3.jpg)
Introduction
• Cyber Physical Systems are distributed systems-of-systems that perform reliable data acquisition in order to build efficient data models.
• Modeling natural enviroments vs engineered systems
• These models can be used for monitoring, tracking and predicting the dynamics of the physical phenomenon
• Data models aid decision-making procedures under resource constraints.
![Page 4: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/4.jpg)
Introduction
• Example: Traffic Scene Modeling
![Page 5: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/5.jpg)
Main Challenges
• Incomplete sensing
• Algorithmic limitations
• Human and social dimension
![Page 6: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/6.jpg)
Problem Description
• Finding components of a scene
• Understanding the relations between components in
a scene
– Insight into cause of existing relations
– Disambiguation
• Predicting the evolution of a scene
![Page 7: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/7.jpg)
Problem Description
• Example: Simple traffic situation
![Page 8: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/8.jpg)
Modeling Methodology
• Constructing ontology description for vehicular traffic applications
• Constructing traffic scene representation
• Predicting traffic dynamics
![Page 9: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/9.jpg)
Ontology Description
• Concepts• Attributes• Basic Semantic Elements:
– Vehicle attributes– Driver’s driving profile– Cluster of vehicles– Cluster attributes– Cluster-level, social behavior– Cluster dynamics– Road conditions– Weather conditions
![Page 10: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/10.jpg)
Ontology Description
• Example: Simple ontology for traffic applications
![Page 11: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/11.jpg)
Ontology Description
• Relations in ontologies for traffic applications– Enabling relation– Is-part relation
![Page 12: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/12.jpg)
Traffic Scene Representation
• Entity Identification
• Relationship Understanding
• Scene Evolution
![Page 13: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/13.jpg)
Predicting traffic dynamics
![Page 14: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/14.jpg)
Predicting traffic dynamics
• Time of separation
where,
• Time of merging
![Page 15: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/15.jpg)
Experimental Results
• Experiments with six scenarios:– Single vehicle: Good/Bad driving conditions– Cluster of vehicles: Good/Bad driving conditions– A vehicle joining a cluster– A vehicle splitting from a cluster
![Page 16: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/16.jpg)
Experimental Results
• Extract scene elements:
– Sound Sensing and Localization
– SVM-based Clustering
![Page 17: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/17.jpg)
Experimental Results
• Spatial Coordinates extracted from DoA estimates
• Sources of Error in Experimental setup, sound source
• Clustering accuracy: 87.5%• Classification accuracy: 100%
![Page 18: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/18.jpg)
Conclusions
• The proposed methodology models the dynamics of traffic scenes, including the participating vehicles, vehicle clusters, attributes and relations of all scene elements, and related events, like cluster merging and splitting.
• The main steps of the methodology find the elements of a scene, identify the relations among the elements, and construct analytical prediction models for the traffic scene dynamics.
• Compared to other methods, this methodology constructs the models online using data from embedded sensors.
![Page 19: Anurag Umbarkar , Shreyas K Rajagopal and Alex Doboli](https://reader035.vdocument.in/reader035/viewer/2022070404/56813c7c550346895da61b63/html5/thumbnails/19.jpg)
Questions? Comments?