Abstract
The disclosed technology introduces a multimodal, natural user interface (NUI) for autonomous, agent-driven camera systems. By leveraging a Camera Agentic Harness and an on-device, lightweight Vision-Language Model, the system allows users to guide camera behavior without navigating complex graphical menus. The interface processes pre-capture voice intents, in-capture spatial interactions, and real-time aesthetic feedback. This enables an AI agent to dynamically orchestrate camera subsystems, such as stabilization, framing, and exposure, which ensures the automated capture aligns seamlessly with the user’s intent, even in dynamic or constrained physical environments.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Picozzi, Google Case Managers; Wang, Wayne; Chan, Louis; Nishimura, Jun; So, Yong Jin Gene; Lin, Elsa; and Chang, Yulin, "Natural User Interface for Agentic Camera", Technical Disclosure Commons, (October 28, 2026)
https://www.tdcommons.org/dpubs_series/12491