Daniel Liu

Torrey Pines Institute For Molecular Studies

Papers

Total Citations

H-Index

About

Daniel Liu is a leading researcher at the intersection of 3D computer vision, adversarial machine learning, and haptic system design. His most impactful work addresses the critical vulnerability of deep neural networks to adversarial attacks on 3D point cloud data—a cornerstone technology for autonomous driving, robotics, and drone navigation. In his highly cited 2020 paper, "Adversarial Shape Perturbations on 3D Point Clouds" (44 citations), Liu pioneered methods for generating subtle, imperceptible shape deformations that can fool state-of-the-art 3D classifiers, revealing fundamental weaknesses in how models perceive geometry. This work has become essential reading for researchers building robust perception systems. Earlier in his career, Liu contributed to the field of haptics with his 2011 simulation model of the Pacinian corpuscle, a mechanoreceptor critical for sensing vibration and texture. This model provided a biophysically grounded framework for designing more realistic tactile feedback in human-computer interfaces, with applications ranging from surgical robotics to mobile devices. Liu’s research uniquely bridges the gap between understanding biological touch and securing the next generation of 3D-aware AI systems.