Cody Houff

Papers

Total Citations

H-Index

About

Cody Houff is a roboticist advancing the frontier of text-guided mobile manipulation and visual sensing for dexterous robots. His research centers on integrating language understanding with physical interaction, enabling robots to interpret natural language commands and execute precise manipulation tasks in unstructured environments. Houff’s major contribution is ForceSight (2024, 6 citations), a pioneering system that uses a text-conditioned vision transformer to predict both kinematic goals—target end-effector poses—and associated force profiles from a single RGBD image. This work bridges the gap between semantic understanding and physical compliance, allowing robots to handle tasks requiring controlled contact, such as pushing or grasping fragile objects. Additionally, his 2023 study on visual contact pressure estimation (4 citations) demonstrates how external cameras can infer grip force from visible gripper deformation, eliminating the need for specialized tactile sensors. This approach makes pressure sensing more accessible for both autonomous and teleoperated systems. Houff’s work is notable for its practical impact: by leveraging vision and language, he reduces hardware complexity while expanding robotic capability. His research is essential reading for students and engineers interested in embodied AI, mobile manipulation, and sensor-free tactile estimation.