Robust Sensing and Perception
In logistics robotics, the sensing and perception of a 3D scene is a basic and important task. The purpose of perception is to segment each individual object, detect their 3D locations and understand the semantic meanings, so that the robotic arm can grasp objects or avoid collisions with them according to the detected 3D locations. Therefore, the locational accuracy directly determines the performance of the whole system. The goal of this program is to develop robust 3D imaging sensors, and superior and time-efficient 2D imaging & 3D point cloud algorithms for robust sensing and perception, including object segmentation and detection, 3D pose estimation of object, point cloud consolidation, etc.
3D Imaging Technology
Novel industrial 3D imaging systems for moving and transparent objects.
3D Imaging Technology Demo
3D Sensing and Perception
Robust, efficient and learning-based 3D visual sensing and perception solutions for logistics scenarios.
Point-Cloud Upsampling and Consolidation
Produce high-quality point-clouds with uniform point distribution and also facilitate high-quality 3D mesh reconstruction.
upsampling output upsampling output
sparse input sparse input
3D Point-Cloud Semantic and Instance Segmentation
Develop algorithms for both semantic and instance segmentation with high accuracy and real-time efficiency.
3D Point-Cloud Semantic and Instance Segmentation Demo