Oculi

Dr. Charbel Rizk, Founder & CTO, Oculi

Prior to founding Oculi, he served in several distinguished leadership and technical roles at Johns Hopkins University, including Associate Research Professor in the Department of Electrical and Computer Engineering (ECE), Principal Professional Staff member, Systems and Lead Engineer, Science and Technology (S&T) Advisor, Innovation Lead, and member of both the S&T Committee and the Innovation Steering Group at the Johns Hopkins University Applied Physics Laboratory (JHU APL).

Throughout his career, Charbel has authored numerous publications, filed a substantial portfolio of intellectual property, and received 12 achievement and recognition awards for technical excellence and innovation.

Widely recognized as a leading innovator, thought leader, and accomplished Principal Investigator, he has established and managed high-impact research and development programs that produced multiple transformative technologies and far-reaching innovations. His proven approach to R&D has consistently accelerated the transition of advanced concepts into operational capabilities.

Charbel was an early pioneer in unmanned aerial vehicle (UAV) technologies, artificial intelligence, machine learning, and autonomous systems. He is also the visionary behind the optimized signal-to-information architecture, a groundbreaking framework that integrates dynamic software-defined multimodal sensing, advanced data fusion, and true edge processing to enable next-generation intelligent systems.

 

Topic: 

Moving the Edge from Imaging to Vision

 

Abstract:

Recent advances in artificial intelligence have dramatically improved machine perception and decision-making capabilities. However, achieving state-of-the-art performance continues to require significant computational and memory resources, resulting in high power consumption, increased latency, and complex system architectures. These challenges are particularly acute in always-on edge applications, where energy efficiency, responsiveness, and security are critical.

In collaboration with GlobalFoundries (GF), Oculi, a fabless spin-out from Johns Hopkins University, is commercializing a groundbreaking computer vision architecture purpose-built for the edge. The technology enables fast, efficient, and secure 2D and 3D vision across virtually any wavelength, delivering a new paradigm for intelligent sensing and processing.

At the heart of this innovation is the Oculi SPU™ (Sensor and Processing Unit), an intelligent, fully programmable vision sensor and the industry's first integrated neuromorphic sensing and processing architecture. Unlike conventional image sensors that continuously stream massive amounts of raw data, the Oculi SPU dynamically extracts and outputs only the most relevant information, significantly reducing downstream processing requirements.

The SPU can be configured in real time to operate in multiple output modes, including traditional images and video, polarity events, intelligent events, and actionable insights. It also supports dynamic adjustment of spatial and temporal resolution, sensitivity, dynamic range, and bit depth, enabling continuous optimization for changing operating conditions and mission requirements.

By combining sensing and processing into a unified architecture, the Oculi SPU fundamentally redefines the efficiency of machine vision systems. Compared to conventional imaging-based approaches, systems powered by the Oculi SPU can reduce the latency-energy product by more than 600× while delivering superior performance at a fraction of the cost. This breakthrough enables a new generation of intelligent edge systems for applications spanning autonomous platforms, industrial automation, defense, security, robotics, and smart infrastructure.

 

Key Technologies Covered:

  • Edge AI Vision Processing for always-on systems with strict power and latency requirements
  • Neuromorphic Sensing and Processing Architecture in a single chip vision platform
  • OCULI SPU (Sensor & Processing Unit) fully programmable intelligent vision sensor
  • IntelliPixel™ In-Sensor (in-pixel) Computing reduces bandwidth, energy consumption, and system complexity
  • Multi-Modal AI Vision Data Output - Dynamically configurable output modes
  • Smart Event-Based Vision Technology captures only meaningful scene changes, reducing redundant data processing
  • Programmable Vision Parameters with real-time adjustments for applications needs
  • 2D and 3D Vision Capability to supports both conventional imaging & depth-aware perception
  • Ultra-Low Power and Low-Latency AI Vision for always-on battery operated application use cases