STMicroelectronics is wiring humanoid vision into the NVIDIA robotics stack

STMicroelectronics and Leopard Imaging introduced a Jetson-ready multimodal vision module for humanoid and advanced robotics systems on March 16, 2026. The module combines 2D imaging, 3D depth sensing, and motion perception with NVIDIA Holoscan Sensor Bridge, Jetson, and Isaac integration.

STMicroelectronics was formed in 1987 through the merger of SGS Microelettronica and Thomson Semiconducteurs and is headquartered in Geneva. In robotics, its role is not to build the robot. It supplies the sensing, microcontroller, motor-control, imaging, and inertial components that make robot perception and motion systems easier to integrate.

Leopard Imaging's role is integration. The California-based camera module specialist packages ST sensors into Jetson-ready vision hardware for robot developers, while ST supplies the core semiconductor components. The release cites ST's VB1940 automotive-grade RGB-IR 5.1-megapixel image sensor, VD55G1 global-shutter image sensors, a VL53L8 time-of-flight sensor, and an LSM6DSV32X inertial measurement unit.

The product is aimed at humanoids and other advanced robots where size, weight, power, synchronization, and integration work can slow development. A robot builder can assemble cameras, depth sensors, IMUs, and software bridges independently, but each interface adds friction. ST and Leopard Imaging are packaging the vision layer so builders can spend less time wiring sensor stacks and more time validating robot behavior.

The competitive field includes RealSense, Orbbec, Stereolabs, Luxonis, Sony image-sensor-based modules, and custom OEM camera stacks. ST's distinction is the semiconductor layer: sensors, MCUs, and motor-control components connected directly into the NVIDIA robotics ecosystem.

The module positions STMicroelectronics as a physical AI component supplier, not a humanoid brand. If humanoid builders converge around Jetson and Isaac workflows, pre-integrated sensing modules become important infrastructure: the robot's perception layer becomes less custom, more repeatable, and closer to production design.

Have a robotics update Korthos should review? Send news, deployments, product releases, funding rounds, research, or media to tips@korthos.xyz or reach out on X at @agkorthos.