Based Estimation

Based estimation, using visual data as the primary input, aims to infer various properties and states from images and videos. Current research focuses on developing robust algorithms, often employing deep neural networks like recurrent video transformers and convolutional neural networks, to estimate parameters such as terrain physical properties (friction, stiffness), human body pose, and even subtle indicators like animal pain or human emotional intensity. These advancements have significant implications for diverse fields, including robotics (autonomous navigation and human-robot interaction), healthcare (monitoring patient engagement and fatigue), and space exploration (autonomous spacecraft navigation).