The AI model works. Perception and navigation have been validated. The robot can move through a controlled environment. The next challenge is no longer proving the concept—it is designing the production system around it.
For teams starting with Arduino® VENTUNO™ Q, this transition is neither a complete restart nor a simple replacement of one board with another. Application logic, AI models, and development workflows may provide continuity. The production hardware instead needs to be engineered around the robot’s actual operating requirements: how many cameras it needs, how actuators are connected, where time-sensitive control functions run, and how power and thermal limits are managed inside the final enclosure.
A production SOM is not a one-to-one replacement for a development platform
VENTUNO Q provides an integrated environment for rapidly developing and validating robotics applications. A production-oriented system-on-module follows a different design logic, with the flexibility to adapt the hardware architecture to the requirements of the final system.
SECO’s SOM-SMARC-Dragonwing-IQ8 combines Qualcomm Dragonwing™ IQ8 Series processing with up to 40 dense TOPS of AI performance in the compact SMARC® Rel. 2.2 form factor. As a production-oriented module, it provides the processing, memory, and high-speed I/O foundation required to build application-specific robotics systems.
Unlike an integrated development platform, a production SOM is designed to be integrated into an application-specific system rather than replicate every connector or subsystem used during prototyping. It is instead integrated through a carrier board designed around the final robot, where application-specific interfaces, control resources and other supporting components can be selected according to the target system.
Moving from an open development setup to a robot operating continuously inside an enclosure introduces another set of practical constraints. Sustained AI workloads must be considered together with power availability, ambient temperature and thermal dissipation. SOM-SMARC-Dragonwing-IQ8 supports passive thermal management and is available in an industrial-temperature configuration covering -40°C to +85°C, giving designers greater flexibility when integrating the module into the final mechanical and thermal architecture.
Designing the carrier around robotics-critical interfaces
The carrier board determines how the compute module interacts with the rest of the robot. Its design should begin with the application architecture rather than with a direct comparison between the I/O maps of two platforms.
For robotics designs, SOM-SMARC-Dragonwing-IQ8 provides the connectivity needed to build around different sensing and control architectures, including configurable PCIe Gen4, dual Ethernet with up to 2.5 GbE, dual CAN-FD, UARTs, and 2-lane and 4-lane MIPI-CSI options for vision systems. The engineering question is not simply which interfaces are available, but which ones the final robot actually needs.
Those requirements shape the carrier architecture itself: vision systems may call for different MIPI-CSI configurations, control and sensing functions need to be selected and mapped across the available CAN-FD and PCIe resources, while network bandwidth and physical connectivity should reflect how the robot will interact with machines and infrastructure in its target environment.
The AI compute platform, control architecture, and physical design of the robot must be considered as one system.
What can remain continuous—and what requires adaptation
Moving to a production architecture does not necessarily mean discarding the application developed during prototyping. AI models, high-level application logic, selected software components, and familiar development workflows may remain reusable, depending on their hardware and software dependencies.
This continuity does not mean that the move to production is a one-click migration.
Software portability therefore needs to be considered together with the carrier architecture and the requirements of the final system. A model may run on the same AI engine, but the surrounding software must still communicate correctly with the cameras, sensors, control systems, and other peripherals selected for the production robot.
SECO supports this transition with carrier board design expertise, I/O mapping, BSP support, driver integration, and embedded engineering services. This helps teams coordinate hardware and software development rather than treating them as separate workstreams.
Partitioning AI and deterministic control
A key architectural decision is how AI processing and time-sensitive control functions should be distributed across the production system. Depending on the application, these functions may rely on different compute and control resources, with latency, isolation, safety and control requirements guiding the most appropriate partitioning.
This becomes particularly important when moving from an integrated prototyping platform to a modular SOM-based architecture, where control functions may need to be redistributed according to the requirements of the final robot.
Validating the platform before the custom carrier
Before moving to a custom carrier, engineering teams can evaluate SOM-SMARC-Dragonwing-IQ8 using SECO’s EVAL-KIT-SOM-SMARC and Modular Vision Kit. These platforms support early module validation, peripheral testing and software integration while the final application-specific carrier is being developed.
They do not replace final-system validation, where the custom carrier, thermal design, control architecture and application-specific interfaces must be verified together in the target application.
Coordinating hardware and software from the beginning
Hardware integration alone does not define the production platform. The software foundation—including the operating system, BSP, drivers and application environment—must evolve together with the target hardware.
Application portability and platform software support are separate parts of the transition. AI models and application logic developed during prototyping may remain reusable, while BSPs, drivers and hardware-specific integration must still be adapted to the target system. The software environment for SOM-SMARC-Dragonwing-IQ8 combines platform-level Linux support with the path provided by Clea OS toward a consistent industrial foundation for application deployment, maintainability and lifecycle management.
Carrier design, BSP development, driver integration, and application porting should therefore progress as coordinated activities.
Designing the platform for more than the first robot
A production architecture should support not only the first deployment, but also future robot variants and product generations.
The modular nature of SMARC allows engineering teams to develop carrier architectures that may be reused across related products, provided that interfaces and system requirements remain compatible. This can reduce future redesign effort and help manufacturers evolve sensing configurations, application software, and compute options without rebuilding the entire platform.
SOM-SMARC-Dragonwing-IQ8 provides the industrial compute foundation. SECO’s carrier board expertise, embedded engineering support, and Clea software infrastructure help teams design the complete robotics platform around it.
Register your interest in evaluating SOM-SMARC-Dragonwing-IQ8 for your next industrial robotics platform.