Following are titles and abstracts for presentations that I can give. Each can be adapted to fit a time slot from 30-60 minutes.
Title 1: Automated Precision Mapping of Roadway Features
Abstract: Next generation roadway maps and vehicle navigation systems for Intelligent Transportation System (ITS) applications (e.g., intersection collision avoidance) have the objective of reliably achieving where-in-lane positioning (decimeter) accuracy. Feasible and cost effective meathds forattaining roadway feature maps (RFMs) with this accuracy are required. This presentation considers RFM construction based on algorithms that automatically process LIDAR, camera, GNSS, and inertial measurement data to map roadway features (e.g., lane edges, signs, curvature, stop lights). The raw sensor data generated is on the order of one terabyte per 30 miles, yielding a big data problem. GIS tools have utility both for the raw data and for the RFM storage, management, and distribution. The presentation will discuss data acquisition, RFM sensor characteristics, mapping algorithms, the role of GIS in both map construction and ITS application support, and will present experimental results from a demonstration at the Turner Fairbanks Research Center that demonstrate mapping and real-time vehicle position estimation at sub-decimeter levels.
Title 2 : Target Tracking in Automated Distributed Camera Networks
Abstract: Despite large quantities of data, the performance of scene understanding algorithms in camera networks often suffers due to the inability to effectively acquire features on targets. This presentation discusses an approach for collaboratively and dynamically controlling the pan, tilt, zoom (PTZ) parameters of a distributed camera network to maximize various scene analysis performance criteria, like tracking and recognition accuracy, through opportunistic acquisition of high quality images. The overall approach requires distributed and cooperative approaches for three processes: feature-to-target association, target state estimation, and PTZ optimization. The presentation will present an introduction and overview of such systems, present a Bayesian-risk based formulation of the PTZ distributed optimization problem with a game-theoretic solution that allows the global problem to be decoupled into local problems, introduce the idea of naivety (limited local target observability) that frequently occurs in distributed camera networks, and present a newly derived information-weighted consensus filter (ICF) that addresses the issue of naivety. We compare the performance of the ICF with existing consensus algorithms analytically, as well as experimentally by considering a distributed camera network under various operating conditions.
Biography: My biography is available at bio link.