UH Mānoa researchers present work on robot perception and learning at CoRL
The Conference on Robot Learning (CoRL) in Austin selected 3PTv2, the successor to the 3D-Object Perception Transformer (3PT), for a main-conference poster. CIRP doctoral student Agastya Kalra and co-authors, including Huaijin Chen, developed 3PTv2 for zero-shot pose estimation. It estimates the poses of objects unseen during training with substantially faster inference and higher accuracy.
CoRL also selected Telescopes are Optomechanical Robots by UH Mānoa researcher Justin Fletcher for a main-conference poster. At the UMI Arena workshop, Yosub Shin, Yuan Liu, and Igor Molybog presented a poster on tracking gloves with a head-mounted camera (P09) and demonstrated their YAM-UMI device (D10).






