Conducting research and development of multi-modal vehicle perception and mapping systems. Specifics include:- Deep learning for scene understanding and autonomous driving: from scene element detection, segmentation and tracking to applications and domain adaption of large multi-modal foundation models and beyond- LiDAR processing: point clustering/classification, object detection, object tracking, 3D environment reconstruction- Sensor fusion: joint processing of GPS signal, map and real-time sensing results for enhanced vehicle perception and positioning

Responsible for algorithm and software prototype development of a synchronized moving grid cone-beam CT system. Specifics include: - Object (Marker) detection of X-ray images- Geometric calibration of the imaging pipeline- Physics-based image enhancement- Exemplar-based local noise modeling

As the technical lead, designed and developed a 3D immersive multi-camera video surveillance system. Key features are:- 3D scene reconstruction using dense laser survey scans- Fast algorithms for large-scale LiDAR point cloud processing and rendering- Real-time fusion of surveillance video feeds and a 3D virtual reality environment (the scene that is monitored)- Interactive immersive navigation in the VR with augmented geospatial awareness of the video contents- Cross-camera object tracking- A distributed video streaming, processing and storage architecture supporting scalable video analyticsAlso, as applicant and the award recipient, attracted 150,000 RMB external funding from Human Resources and Social Security Bureau of Beijing, China(A merit-based funding program for scientific and technological activities of persons with overseas experience, 2012.)

Investigated field laser scan mechanisms and large-scale LiDAR data processing, surface reconstruction and efficient rendering in a State MTTC project "HyperEye: Susceptibility Weighted Imaging-based Informatics Tools for Brain Tumor Studies". Other roles include 1) Writing industrial and governmental grant proposals; 2) Advising junior master and doctoral students, helping them develop their research topics and guide them through project challenges.

Conducted funded research on computational geometry and digital geometry processing at the Department of Mathematics. Achievements include:- A theoretically strict area-preserving surface parameterization technique using Lie advection- Intrinsic geometric scale space (IGSS) using 2-manifold Ricci flow, with application to scale-invariant geometric feature extraction- 3D Merger: A software used to merge 3D object scans from different perspectives into a whole piece; Core techniques include ICP point-cloud registration, distance field computation and iso-surface extraction- A method and prototype software to automatically paint 3D face models with separately captured photos using least squares conformal maps.

As team member, participated in a number of NSF and NIH funded multi-disciplinary research projects; conducted independent research on computer graphics, geometric modeling and visualization, with applications to brain imaging data analysis and analytics; coordinated teamwork with partners from medical school for clinical validation. Contributed software modules and technical methods include:- Landmark-constrained deformable surface registration based on conformal geometry- Salient geometric feature detection via geodesic scale space- Fast tetrahedral mesh (a discrete 3D volume representation) parameterization using 3-manifold harmonic mapping- Multimodal data integration based on a conformal brain model, which allows quantitative assessment of structural and functional changes with complementary neuroimaging technologies during the progression of epilepsy

Participated in the development of an Intel multi-CPU-based HDTV system: - Performance optimization on computation-intensive modules, e.g., motion estimation, image quality assessment, etc.- Software QA testing