Jingnan Shi

Jingnan Shi is a PhD candidate whose research is focused on perception and vision-based navigation for robots and autonomous systems with the goal of improving the safety, durability, and effectiveness of these systems. As a MathWorks Fellow, Jingnan will pursue three primary projects: development of a new algorithm to deliver fast and certifiably robust 3-D perception, a tool that has been successfully applied in several large-scale robotic systems and shared with the research community; a graph theoretic framework for outlier rejection in various geometric perception problems, offering greater speed and robustness than existing solvers; and category level perception for self-driving vehicles, involving the creation of a novel solver for estimating shapes and poses of cars surrounding an experimental robot. His plans include designing new methods to incorporate learned robustness into robotic perception systems, including the investigation of differential programming to improve robot robustness through unsupervised learning. MATLAB is key to Jingnan’s work, enabling him to bridge grounded algorithms with high-performance implementations and to make valuable contributions to the field of robotics.