• Heng Yang and Luca Carlone, In Perfect Shape: Certifiably Optimal 3D Shape Reconstruction from 2D Landmarks, In IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2020. —We present Shape*, the first certifiably optimal non-minimal solver for 3D shape reconstruction from 2D landmarks in a single image, and Shape#, a robust shape reconstruction algorithm that tolerates a large amount of outliers in the 2D measurements. We demonstrate the power of Sums-of-Squares (SOS) relaxations in solving challenging geometric perception.

  • Heng Yang, Pasquale Antonante, Vasileios Tzoumas and Luca Carlone, Graduated Non-Convexity for Robust Spatial Perception: From Non-Minimal Solvers to Global Outlier Rejection, IEEE Robotics and Automation Letters (RA-L), 2020.—A general-purpose robust estimation framework called graduated non-convexity (GNC) that can be applied to any problem where a globally optimal non-minimal solver (e.g. from SDP and SOS relaxations) for the outlier-free case is available. Empirically robust againts 70-80% outliers across applications. As an alternative to RANSAC, GNC robustifies non-minimal solvers, while RANSAC robustifies minimal solvers.



  • Heng Yang and Luca Carlone, A Quaternion-based Certifiably Optimal Solution to the Wahba Problem with Outliers, International Conference on Computer Vision (ICCV), 2019. Oral presentation (4% acceptance rate) (Video—Introducing QUASAR (QUAternion-based Semidefinite relAxation for Robust alignment), a quaternion-based SDP relaxation that is tight and returns the certifiable global optima to the Wahba problem (aka rotation search) in the presence of large noise and 95% outliers.



  • Heng Yang, Man M Nguyen, Sheng-Wen Huang, Vijay Shamdasani, Hua Xie and Brian W Anthony, Simulation, design, and implementation of external mechanical vibration for ultrasound shear wave elastography. 2017 IEEE International Ultrasonics Symposium (IUS). — We propose to induce shear waves using a multi-point-source external mechanical vibration concept. Simulation and experimental prototype show promising results on liver phantoms.
  • Julian Chacon-Castano, Daniel R Rathbone, Rachel Hoffman, Heng Yang, Dimitrios Pantazis, Jason Yang, Erik Hornberger and Nevan C Hanumara, Music and the brain-design of an MEG compatible piano2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). — A piano design by optical fiber encoder that can be used in MEG room to study human brian activity induced by playing music.