A robotics edge-compute development kit based on NVIDIA's Blackwell architecture.
AI Hardware
Compute platforms and components built to run perception, reasoning, or control close to the physical world.
v0.1 inclusion rule
Must materially enable on-device or edge intelligence rather than being general-purpose compute with an AI label.
An add-on board that gives a Raspberry Pi 5 a Hailo-10H AI accelerator and 8 GB of its own memory, so it can run camera vision models, and small language models, on the board itself.
A robotics development kit built on AMD's Kria K26 module, which pairs Arm processor cores with programmable FPGA logic so builders can run perception and control pipelines in hardware, with native ROS 2 support and industrial networking.
A stereo depth camera with its own 52 TOPS processor that runs detection and depth models on the camera itself, so a robot or machine can get perception without a separate computer.
A development kit for ST's STM32N6, a microcontroller with its own neural accelerator, so builders can run camera and audio AI models at microcontroller scale and power.
Qualcomm's robotics and IoT development kit on its QCS6490 processor, rated at 12 TOPS, with Qualcomm's robotics software for mapping, navigation and object detection. Thundercomm builds and sells it.