Roger Tsai published 26 papers during the one and a half year alone at University of Illinois at Urbana Champaign to earn him his PhD degree in Electrical Engineering and Computer Science, followed by his tenure at IBM for 27 years. He was known for his pioneer work on 3D Robotics Vision. His work on relating Singular Value Decomposition to the properties and structures of relationships between 2D images and 3D motions is one among the seminal work building up the foundation of 3D Machine Vision for the field. On Caltech's J. Bouguet's well known list of camera calibration links, Tsai Camera Calibration is labeled as "A classic! Everyone interested in camera calibration should know about that link". Google Scholar reported more than 6000 citations of his work. His work has been used and taught widely worldwide among universities, industries and commercial practices, such as Callaway Golf Company for analyzing golf ball flight trajectories, Hunter Engineering Corp for measuring wheel alignment, MIT Media Lab for projects, to name a few. As a result, he took up leadership in the professional societies such as Chairman of the Computer Vision Committee, IEEE Society for Robotics and Automation; Technical Editor, Robotics Vision, IEEE Transaction on Robotics and Automation; Editorial Board, Robotics Review, MIT Press; Advisory board, International Societies of SPIE Sensor Fusion, Cambridge Symposium, and Chairman, Risk Management and Demand Forecasting, INFORMS conference, to name a few. He published 20 journal papers and book chapters, and some 113 publications.
In the middle of his career at IBM, he began to switch field into business analytics, manufacturing process optimization, data mining, focusing more on IBM's proprietary business improvement rather than external academic publications. He used his considerable analytic inclination in a wide range of disciplines to help IBM reengineer numerous corporate, manufacturing, consulting services, sales and business diagnostics and practices to enhance the quality, capability and performances in many fronts, resulting in higher financial returns. His optimization and statistical analysis work helped IBM become the first in the industry to use copper for back-end-of-line semiconductor processes, which revolutionalize semiconductor manufacturing, and the first to produce high density disk read-write head for laptops. His multivariate statistical optimization technique was key in IBM's ability to produce the industry-leading dilute chemistry web clean process.
Overall, more than 20 manufacturing processes were improved. His statistical forecasting, data mining and predictive modeling helped IBM significantly improve the prior forecasting and planning effectiveness, and helped IBM take proactive steps in supply/demand matching and manufacturing optimization. His analytic, algorithmic, data mining and predictive modeling also helped IBM improve several inventory planning and statistical optimization consultation services and products, and helped decision making senior executives in understanding business strengths and weaknesses in order to take the proper action in sales and business unit planning. In 2010, he started his teaching/research career at Fordham University, fulfilling his long time dream and aspiration to teach and to help young people excel, leveraging his considerate industry experience and his love for teaching. He now teaches web programming, discrete structures for Computer Science, Discrete Mathematics. He is coming back to his old field of robotics vision, collaborating with the existing robotics group at Fordham. He will also pursue his research interest in business intelligence and data mining.