8devices has opened early access for the 8Sight T100, a compact thermal camera integrating a Lynred ATI320 sensor with an STMicro STM32N6 microcontroller. The device is designed for edge AI processing in industrial and robotics applications, supporting H.264 and MJPEG streaming.
The integration of a dedicated NPU and hardware ISP within a thermal imaging unit marks a shift toward localized, low-latency video analytics for specialized industrial streaming. By processing H.264 streams and AI models entirely on-device, the unit bypasses the bandwidth and security constraints of cloud-based thermal monitoring. This development aligns with the broader industry trend of moving computer vision tasks to the edge to support autonomous robotics and UAS operations. As 8devices prepares for a 2027 production run, the market should monitor the release of their RobonodeVision system to see how it simplifies multi-camera integration for complex vision pipelines.
8devices positions the 8Sight T100 as a complete, production-ready thermal camera rather than a development board, distinguishing it from STM32N6-based platforms like the OpenMV N6 and Ohm Lab Neuro N6 that require additional integration work. The company's RobonodeVision system, expected to enter early access in Q4 2026, is designed to simplify multi-camera integration for robotics and UAS vision pipelines, with the 8Sight T100 serving as a thermal imaging node within that ecosystem. 8devices has not yet published an SDK, API reference, or sample code for the T100, instead offering custom firmware and software development services to early-access participants.
Lynred launched the ATI320 as its first thermal imager with embedded image signal processing, targeting camera makers in industrial, consumer, and safety applications who need compact, low-power thermal modules. The sensor's shutterless design provides continuous image viewing without periodic calibration interrupts, a feature Lynred highlights as critical for security-surveillance and firefighting use cases. The ATI320 is available in two variants: the ATI320S (sensor only, under 3 grams) and the ATI320L (with integrated lens, under 7 grams), both meeting Mil-Std-810/883 ruggedization standards. 8devices selected the ATI320 for the 8Sight T100 to leverage this embedded ISP capability alongside the STM32N6's Neural-ART NPU, creating a dual-processing architecture where the sensor handles raw thermal correction and the MCU runs AI inference.
The STM32N6 microcontroller from STMicroelectronics features an Arm Cortex-M55 core running up to 800 MHz with a 600 GOPS Neural-ART accelerator, representing STMicro's push into edge AI workloads previously handled by higher-power application processors. The 8Sight T100's 38-gram weight and 28.5 x 30.5 x 29.5 mm dimensions place it in the same form-factor class as other compact LWIR modules, such as the ISM-ATI320 interface board that pairs the same Lynred ATI320 sensor with NVIDIA Jetson platforms via MIPI-CSI2. That module targets developers who need GPU-accelerated thermal processing, whereas 8devices differentiates the 8Sight T100 by running inference entirely on the microcontroller without requiring a separate compute board, reducing system complexity for robotics and UAS integrators who prioritize weight and power budgets.
The 8devices 8Sight T100 is a new thermal camera that integrates a Lynred ATI320 sensor with an STM32N6 microcontroller. By performing H.264 streaming and AI inference locally, it enables low-latency thermal analysis for robotics and UAS. This shift to edge processing eliminates reliance on cloud infrastructure, improving security and bandwidth efficiency.
The 8Sight T100 uses a dual-processing architecture where the sensor handles thermal correction and the STM32N6 microcontroller runs AI inference locally. This allows for real-time, low-latency video analytics without needing a separate compute board or cloud connection.
Production units of the 8Sight T100 are scheduled for release in Q1 2027, following the availability of engineering samples in late 2026.
The 8Sight T100 is a compact device weighing 38 grams, with dimensions of 28.5 x 30.5 x 29.5 mm, making it suitable for integration into robotics and unmanned aerial systems (UAS).
No, the 8Sight T100 is designed as a complete, production-ready thermal camera that runs AI inference entirely on its internal microcontroller, distinguishing it from modules that require external platforms like NVIDIA Jetson.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source