Industry Event
Vision Tech AI Session: Sensor-Optics Co-Design for AI Vision Systems: A Systems-Level Approach to Application-Tuned Imaging Performance
21 January 2026 • 11:30 AM - 12:00 PM PST | West Expo Stage 1 (Moscone West, Exhibit Level)
Artificial intelligence (AI) vision systems have emerged as critical enablers across a broad spectrum of applications—from autonomous vehicles and smart cities to consumer electronics, industrial robotics, and medical diagnostics. The demand for real-time, high-fidelity visual data has escalated the performance expectations of embedded vision systems, particularly those operating at the edge with constrained power, processing, and thermal budgets.
However, a persistent challenge across many vision system deployments is the performance gap between lab-tested AI models and real-world behavior. In many cases, this disparity can be traced not to the algorithmic components of the system, but rather to foundational hardware design decisions—particularly the mismatch between image sensor architecture and the optical system.
This session explores the necessity of sensor-optics co-design as a fundamental engineering methodology for the development of next-generation AI vision platforms. We argue that siloed component selection—treating sensors and optics as interchangeable or modular elements—often leads to significant performance degradation in deployed systems. Instead, we propose a systems-level co-design approach, in which optical parameters such as MTF, F-number, field of view, and lens aberrations are jointly optimized with sensor parameters like quantum efficiency, dynamic range, pixel size, and shutter type.
With insights derived from field-deployable imaging solutions at Optopax, we present a framework for designing imaging pipelines that are robust, efficient, and application-specific.
MENU: Coffee, decaf, and tea will be available nearby.
SETUP: Theater seating.
However, a persistent challenge across many vision system deployments is the performance gap between lab-tested AI models and real-world behavior. In many cases, this disparity can be traced not to the algorithmic components of the system, but rather to foundational hardware design decisions—particularly the mismatch between image sensor architecture and the optical system.
This session explores the necessity of sensor-optics co-design as a fundamental engineering methodology for the development of next-generation AI vision platforms. We argue that siloed component selection—treating sensors and optics as interchangeable or modular elements—often leads to significant performance degradation in deployed systems. Instead, we propose a systems-level co-design approach, in which optical parameters such as MTF, F-number, field of view, and lens aberrations are jointly optimized with sensor parameters like quantum efficiency, dynamic range, pixel size, and shutter type.
With insights derived from field-deployable imaging solutions at Optopax, we present a framework for designing imaging pipelines that are robust, efficient, and application-specific.
Speaker
 |
|
Event Details
FORMAT: Oral presentation followed by audience Q&A.MENU: Coffee, decaf, and tea will be available nearby.
SETUP: Theater seating.