SBIR-STTR Award

This Small Business Innovation Research Phase I project develops simultaneous localization and mapping (SLAM)-based innovative methods for geo-registration and demonstrates the feasibility and effectiveness of the methods for accurate geo-registration of wide area motion imagery (WAMI). The proposed methods are capable of dealing with noisy sensor metadata and do not rely on the availability of accurate and/or up-to-date geo-registered reference data. The key innovations in this effort are: i) development of efficient incremental SLAM based geo-registration to enable large scale estimation, ii) incorporation of visual odometry priors to handle non-smooth motion, iii) incorporation of priors from reference data for geographically consistent mapping and error handling, and iv) development of a distributed framework to enable airborne processing in SWaP constrained environment. The project further offers analysis of the algorithms with respect to a wide variety of system parameters, such as, types of pose constraints and their modeling, choice of features, choice of data association schemes, etc. The Phase I effort will include: development of enabling algorithms, implementation of a video geo-registration system, demonstration of proof of concept, and evaluation and failure mode analysis of the proposed technologies using real WAMI data.

Benefit:
Wide-area motion imagery has proven to be a critical asset for persistent surveillance and reconnaissance of large geospatial regions. However the utility and value of these assets to the analysts significantly depends on the quality of geo-registration. For example, the performance of common applications such as targeting, tracking of high valued targets, and activity analysis, all are a function of georegistration accuracy of the data. Hence the development of robust geo-registration technologies is critical for enabling truly automated real-time wide area surveillance and reconnaissance. The proposed technologies facilitate accurate geo-registration on large scale imagery in the presence of sensor noise and errors in geo-registered reference data while overcoming fundamental and operational challenges and enable many automated tools for the exploitation of WAMI data. These include: • Tracking and handoff of high-value targets to and from WAMI data. • Precision targeting. • Mission Planning • Automated Activity Analysis • Real-time monitoring, and • Forensic data analysis.

Keywords:
Geo-Registration, Simultaneous Localization And Mapping (Slam), Visual Odometry, Incremental Slam, Wide-Area Motion Imagery, Distributed Slam.

This Small Business Innovation Research Phase II project will develop innovative simultaneous localization and mapping (SLAM)-based methods for accurate geo-registration of wide area motion imagery (WAMI). The proposed computational architecture and key technologies are capable of handling noisy sensor metadata, and scalable to process large-scale WAMI data in real-time using onboard, distributed processing units such as field-programmable gate arrays (FPGA). The key innovations in this effort are: i) development of distributed SLAM based geo-registration with centralized fusion to enable large-scale WAMI processing, ii) development of computationally efficient and FPGA-suitable feature matching approaches, iii) development of fast camera ray-to-3D terrain model intersection algorithm, and iv) development of a distributed computational framework to enable airborne processing in SWaP constrained environment. The Phase I effort demonstrated the proof of concept by developing core technologies that form the basis of the proposed algorithms. The Phase II effort will be focused towards advancement and integration of these enabling technologies and will include development of new algorithms to handle the issues identified during Phase I, transitioning of technology and prototyping, integration of technologies into an existing ISR system, detailed quantitative and qualitative evaluation at component and system levels, and demonstration of technologies in operationally representative scenarios.

Benefit:
Wide-area motion imagery has proven to be a critical asset for persistent surveillance and reconnaissance of large geospatial regions. However the utility and value of these assets to the analysts significantly depends on the quality of geo-registration. For example, the performance of common applications such as targeting, tracking of high valued targets, and activity analysis, all are a function of georegistration accuracy of the data. Hence the development of robust geo-registration technologies is critical for enabling truly automated real-time wide area surveillance and reconnaissance. The proposed technologies facilitate accurate geo-registration on large scale imagery in the presence of sensor noise and errors in geo-registered reference data while overcoming fundamental and operational challenges and enable many automated tools for the exploitation of WAMI data. These include: ? Tracking and handoff of high-value targets to and from WAMI data. ? Precision targeting. ? Mission Planning ? Automated Activity Analysis ? Real-time monitoring, and ? Forensic data analysis

Keywords:
Geo-Registration, Wide-Area Motion Imagery, Simultaneous Localization And Mapping (Slam), Distributed Slam, Geo-Referenced Data, Feature Matching, Sensor Fusion