NXP Unified Robotics Platform Accelerates Physical AI Deployments

NXP Semiconductors N.V. has unveiled a unified robotics platform that combines hardware and software with NXP's secure, safety-enabled technologies, initially targeting the drone market. The modular, open-source software development kit (SDK) and reference designs, built on a neural axis architecture, connect sensing, real-time control, AI processing, and communications across robotic systems.

The open robotics SDK provides full-stack robotics enablement in one development foundation with validated software that is maintained and continuously updated. It provides a single entry point across all NXP robotics platforms.

The unified platform simplifies integration with software and pre-validated, interoperable subsystems that scale from individual functions to complete systems. It can help developers build systems faster, integrating functionality such as AI, connectivity, motor control, security, and energy management. It can reduce development time from months to weeks.

The intent of this platform is to speed up the deployment of physical AI, said Jerome Legros, vice president and general manager, robotics, at NXP Semiconductors.

NXP will help developers enable subsystems by releasing reference designs with validated software, based on the modular and open-source SDK, and ensure that all the subsystems, such as compute and flight management, are pre-validated together, Legros explained. "The subsystems are fully interoperable so that it allows customers to easily scale from an individual function to a complete system."

Underpinning this platform is NXP's focus on three "non-negotiable" key performance indicators (KPIs): trust (safety and security), latency (real-time performance), and power, which are requirements to scale, according to Legros. These align with NXP's neural axis architecture for robotics, consisting of reasoning, coordination, and reflexive layers, which is scalable across all form factors.

"We believe that the intelligence needs to be distributed in the system, which is what we call the neural axis, where you have three layers [cerebrum/reasoning, cerebellum/coordination, and spinal cord/reflexive] in any robotic system," Legros said.

There is the cerebrum, similar to a human being, which is the reasoning layer that plans the next move; the cerebellum, which is the coordination layer, taking care of all the actuation of the body and also safety and ensuring the stability of the drone; and the spinal cord, which is the reflexive layer, and this architecture is common across all form factors of robotics, he explained.

Legros said NXP has a unique position in the industry, supporting all three layers with scalable hardware, safety and security, motor control, connectivity, and networking capabilities.

Addressing customer pain points

The platform is designed around the need to provide a single validated development experience that developers can scale from single functions to complete trusted physical AI systems. This helps them develop products faster with pre-integrated software and reference designs and reduce complexity with validated subsystems that are interoperable.

The platform addresses three key customer pain points: higher system complexity due to AI and technology integration, and the need to accelerate development and keep devices safe and secure during their lifetime. "Complexity is increasing; we need to tackle the new requirement, which is trust; and development has to go faster to shrink time to market, and we are in an environment-robotics, which is going at a very high speed," Legros said.

"When you look at all these requirements, this is why we need to shift from a component play to a system play, and this is what we offer to our customers," he added. "We want to give them a solution which is simple and fast to help with their time to market."

But due to a very fragmented market, including in robotics, Legros does not expect one reference design for a drone, as an example, to fit all needs. "We believe that flexibility is very important, and for any subsystem or reference design we are always focusing on the three KPIs, which is power, latency, and trust in our design."

NXP cited an example in which a developer may begin by evaluating a motor control subsystem, which has been tested and validated with the SDK. Using the unified robotics SDK, reference designs, and platform architecture, developers can then add AI-based perception, flight control, and navigation to build a complete autonomous drone faster.

One of the challenges that developers face is enabling robotic subsystems to work together across every form factor. This can take months to integrate and validate, Legros said. The robotics platform cuts this effort to weeks by providing pre-integrated software and validated subsystem interoperability, built on NXP's portfolio, he added.

Today, customers must select the hardware, such as compute, motor control, flight management, or networking, then download the SDK, validate the SDK on the hardware, and make sure that the system is working, Legros continued.

The next step is to integrate all these subsystems together, validate them, and then test the solution, he added. "It's a lot of manual integration."

There may be late discovery and the developer will need to go back and download a new SDK to fix the issue, so the cycle to integrate and reach the first launch is long, which is what NXP is helping to resolve, he added.

NXP also helps to maintain and update the software in these products. NXP makes sure all the subsystems are validated together, so the customer is getting one package in a validated baseline with lower development risk, Legros said. "The system is going to be maintained in terms of software and the system KPIs are optimized."

The unified robotics platform is designed to work with industry-standard robotics frameworks and delivers a common hardware and software foundation for robotics development. The open-source, modular SDK provides drivers, middleware integration, reference applications, configuration tools, and reusable software components. This includes integrations with open-source technologies including PX4, Zephyr, and ROS 2. Legros said NXP will create a GitHub repository for its open-source full-stack robotics enablement.

Reference designs

The platform's prototype-ready reference designs incorporate secure processing, deterministic real-time control, and safety-ready system architectures built from NXP's hardware portfolio.

The first reference designs support drones, including autonomous compute and flight control using the NavQ95 based on the i.MX 95 applications processor; flight management using the Tropic VMU/FMU based on the i.MX RT1064 crossover microcontroller (MCU); intelligent sensor networking using the 100Base-T1 sensor node based on the MCX N947 MCU; and motor control using the AM32-compliant electronic speed controller based on the MCX A13 MCU.

Each reference design includes the safety and security enablement of the design's primary processor. It can be enabled independently of robotics-specific functionality. According to NXP, developers can enable an initial drone functionality in 30 minutes out of the box, using a reference design supported by the robotics SDK.

The platform will extend beyond drones across a variety of autonomous systems that require perception, navigation, motor control, connectivity, and safe real-time operation. These include service robots, autonomous mobile robots, and other autonomous systems.

The unified robotics platform is available now, including reference designs for drones and the supporting SDK. Reference designs for other autonomous systems are expected to be available in in the first half 2027.

NXP Semiconductors is exhibiting at electronica 2026 in Munich, Germany, November 10-13, at Hall C5, Booth 179.

The post NXP Unified Robotics Platform Accelerates Physical AI Deployments appeared first on Embedded.