Physical AI & Humanoid Robotics: NXP + NVIDIA Holoscan Sensor Bridge Reference Designs

In this EdgeVerse TechCast episode, hosts Kyle Dando and Bridgette Stone speak with NXP's Altaf Hussain about Physical AI and a new NXP–NVIDIA collaboration delivering integrated, real-time robot body solutions built around NVIDIA's Holoscan Sensor Bridge (HSB). They explain how Physical AI differs from cloud AI by requiring deterministic sensing, decision, and actuation under strict latency and safety constraints, and how NXP's real-time control and networking complements NVIDIA's AI compute. The collaboration launches with two HSB-ready reference designs targeting the hardest robotics problems: perception and motion—an i.MX 95 machine vision design with 10Gb low-latency data paths to the NVIDIA pipeline, and an i.MX RT 1180 distributed motor control design supporting EtherCAT and TSN for synchronized control. Reference software will be published on GitHub.

Episode Resources:

• https://www.nxp.com/hsb-solutions
• https://www.nvidia.com/en-us/technologies/holoscan-sensor-bridge/
• https://docs.nvidia.com/holoscan/sensor-bridge/latest/introduction.html

00:00 Welcome and Topic Setup

00:39 Why Physical AI Is Hard

01:03 NXP and NVIDIA Collaboration

01:59 Defining Physical AI

03:20 Holoscan Sensor Bridge at the Edge

05:27 Two Reference Designs

07:00 Machine Vision i.MX 95

07:43 Motor Control i.MX RT1180

09:19 Software Enablement and Roadmap

10:53 GitHub Access and Wrap Up

11:48 Final Thanks and Next Steps