Chang Liu

Papers

Total Citations

H-Index

About

Chang Liu is an emerging researcher at the intersection of robotics, computer vision, and neural scene representation. Their work centers on the challenging problem of robotic object grasping — developing intelligent systems capable of manipulating objects across diverse shapes, materials, and textures in real-world environments. Liu's most notable contribution is **RGBGrasp**, a pioneering framework that leverages Neural Radiance Fields (NeRF) to enable image-based robotic grasping by dynamically capturing multiple viewpoints during robot arm movement. This work represents a meaningful departure from conventional approaches that rely heavily on expensive point-cloud cameras or large RGB datasets to reconstruct 3D scene information. By harnessing NeRF's powerful implicit scene representation, Liu's method demonstrates that rich 3D understanding can be achieved with more accessible, standard RGB imagery — lowering the barrier to practical robotic deployment. The RGBGrasp framework has accumulated 26 citations across its iterations, reflecting growing interest from the robotics and computer vision communities. Liu's research addresses a fundamental bottleneck in robot manipulation, with implications for automation, assistive robotics, and industrial applications. Their work exemplifies how advances in neural rendering can meaningfully translate into real-world robotic capabilities, making them a researcher worth following as the field rapidly evolves.