Israeli chipmaker Hailo has brought its edge AI technology to CES 2024, demonstrating applications in smart devices by some of the company’s customers.

Its Hailo-8 AI accelerators and Hailo-15 AI vision processors are designed with an “AI-first” approach to perform data center-class AI processing directly on edge devices, improving inference performance at high accuracy and low latency while reducing power consumption.

“The market for edge AI is accelerating with the demand for highly accurate real-time video analytics,” said Orr Danon, CEO of Hailo. “Hailo continues to push the boundaries of AI processing by enabling its customers to embed high-performance deep learning into edge devices. With the recent introduction of Hailo-15 and the expansion of our Hailo-8 portfolio, we’re ready to show CES the disruptive power of edge AI.”

At CES 2024, Hailo is demonstrating edge AI processing in three key markets: video management systems, advanced driver assistance systems (ADAS) and autonomous driving (AD), and security. It is showing the first examples of Hailo-15 AI-powered smart cameras by Truen and Gunsens, as well as multiple security, video management systems, and automotive solutions empowered by Hailo-8 AI accelerators.

For mobility applications, Edge AI is critical to the safety and comfort applications of AD and ADAS, which rely on AI to perceive data from cameras and other sensors for full situational awareness. Hailo says its automotive-grade AI processors provide the power efficiency needed to pack more AI processing into a vehicle and to scale AI compute power to the required level of autonomy—all at an attractive cost for mass-market adoption.

At CES, Hailo’s partner Tier IV is demonstrating a multi-camera detection and segmentation solution using the Hailo-8 AI accelerator. In conjunction with the Renesas smart front camera demonstration, the company is demonstrating high-accuracy object detection from six 2MP cameras processed by the Renesas R-Car V4H Series and Hailo-8 AI accelerator. Also on display is a Hailo-powered ADAS domain controller from iMotion and ECU for off-highway vehicles solution by TT Control.

Pittsburgh-based Velo.ai is showing its Copilot solution integrating Hailo processors for AI-powered electric and conventional pedal bikes. Copilot senses the roadway, looking for approaching vehicles and understanding behavior. Using AI, it can understand when a vehicle safely negotiates passing a cyclist, but it can also identify an aggressive or distracted driver approaching dangerously. In those cases, Copilot alerts the cyclist and the driver to help prevent crashes before they can occur.

The Hailo-15 family of high-performance vision processors is designed to be integrated into intelligent cameras to deliver video processing and analytics at the edge. By introducing AI capabilities into cameras, Hailo says it is addressing a growing demand in the market for better, faster, and smarter video systems. The cameras deliver both AI-powered video enhancement and AI-powered video analytics due to their high compute power, memory efficiency, and AI-focused architecture. They can identify smaller and more distant objects with accuracy and fewer false alarms.

Hailo says its Hailo-8 AI accelerators have been deployed in hundreds of programs and products to date. To meet growing demand, the company recently expanded its offerings to include the high-performance Hailo-8 Century PCIe card line, which offers up to 208 TOPS for more demanding applications, and the Hailo-8L, which makes advanced AI processing available for entry-level applications.