Human Movement Prediction through 2D Imaging:
| This project focuses on methods to determine and use context to predict how humans will behave when working alongside robots. This foreknowledge enables the robotic system to behave safely, prevent collisions, and better support human users. Context is derived from objects visible in the scene, combining results from 2D joint pose estimation and objection detection into a single combined representation. This representation is then used to predict future states through deep learning. |