Skip to main content
SECO

Robotics and Physical AI

Embedded compute, Edge AI and lifecycle support for AMRs, cobots, drones and intelligent machines, from prototype to deployed fleet.

The challenge

Turn Physical AI prototypes into deployable machines

Robotic systems must process multimodal data locally, operate through limited connectivity and scale from a working prototype to a maintainable industrial fleet.

  1. Adaptive intelligence

    Run perception, sensor fusion and inference across heterogeneous compute architectures.

  2. Local AI processing

    Keep latency-sensitive workloads close to sensors and machines when cloud connectivity is limited or unavailable.

  3. Industrialisation

    Move beyond the demo with production hardware, stable software, long-term availability and fleet lifecycle management.

Robotics and intelligent machine applications

Match the compute architecture to each machine’s perception, control, power and lifecycle requirements.

  • MobileAMRs, AGVs and dronesRun perception, sensor fusion and on-device inference within the power and thermal limits of mobile systems.SOM-SMARC-QCS6490 · SOM-SMARC-Dragonwing-IQ8
  • ManufacturingCobots and intelligent machinesSupport perception, sensor fusion and on-device inference for collaborative robots and intelligent industrial equipment.SOM-SMARC-Dragonwing-IQ8
  • VisionInspection and machine visionProcess AI-intensive vision workloads at the Edge with up to 45 TOPS on COM Express Type 6.SOM-COMe-CT6-Dragonwing-IQ-X
  • High computeHumanoids and advanced Physical AI platformsUse high-density compute from Intel® Core™ Ultra and AMD Ryzen™ AI Embedded processors for demanding robotics workloads.SOM-COMe-BT6-PTL · SOM-COMe-CT6-P100

Robotics and Physical AI requirements

Run AI on the machine
Embedded platforms with integrated AI acceleration support perception, inference and decision-making on the device, with performance-per-watt options for mobile and compact systems.
Move from prototype to production
Modular compute, carrier-board design, lifecycle planning and engineering support create a structured path from evaluation to series production.
Match compute to the machine
SMARC and COM Express modules span Qualcomm® Dragonwing™, Intel® Core™ Ultra, AMD Ryzen™ Embedded and Intel Atom® processors, supporting different performance, power and form-factor requirements.
Manage deployed fleets
Clea supports device provisioning, monitoring, staged OTA updates, software traceability and AI workload deployment across distributed fleets.
Cybersecurity for CRA and RED
Clea OS supports secure boot, signed updates and encrypted OTA update mechanisms. The Cyber Security Package supports vulnerability management and compliance paths for CRA and RED requirements.Explore the Cyber Security Package

Recommended products

View all products
  • Edge AI compute for robotics
    New

    SOM-SMARC-Dragonwing-IQ8

    SMARC® Rel. 2.2 module with Qualcomm® Dragonwing™ IQ8 Applications Processors

    • Arm
    • Qualcomm processor
    Processor
    Qualcomm
    Memory
    Up to 32 GB Soldered down
  • Low-power vision and sensor processing
    Available

    SOM-SMARC-QCS6490

    SMARC® 2.1.1 module powered by Qualcomm® Dragonwing™ QCS6490 processor

    • Arm
    • Qualcomm processor
    Processor
    Qualcomm
    Memory
    Up to 16 GB Soldered down
  • High-performance Vision AI
    Available

    SOM-COMe-CT6-Dragonwing-IQ-X

    COM Express 3.1 Type 6 Compact modules with Qualcomm® Dragonwing™ IQ-X Series of processors in SiP-A, featuring a Qualcomm® Hexagon™ NPU with up to 45 TOPS

    • Arm
    • Qualcomm processor
    Processor
    Qualcomm
    Memory
    Up to 64 GB Soldered down
    Edge AI
    Up to 45 TOPS
  • High-performance robotics compute
    New

    SOM-COMe-BT6-PTL

    COM Express Rel. 3.1 Type 6 Module with Intel® Core™ Ultra Series 3 Processors (Codename: Panther Lake-H)

    • x86
    • Intel processor
    Processor
    Intel
    Memory
    Up to 96 GB SO-DIMM
    Temperature
    0°C to +60°C
Clea

Manage robotics fleets throughout their lifecycle.

Clea provisions each device, monitors device and application data, deploys firmware, OS, application and AI model updates in staged rollouts, and records the version running on every unit.

Explore Clea
Provisioning
Assign each machine an identity and configuration profile before deployment.
Telemetry
Monitor device status, sensor health and inference metrics across the deployed fleet.
Updates
Deploy staged, reversible updates to firmware, OS, applications and AI models with a per-device audit trail.
Security
Use secure boot, signed artefacts and CVE patching to protect connected machines throughout their lifecycle.

From Evaluation kit to production

  1. 01Scope the machineDefine sensors, interfaces, AI workloads, control loops, power, thermal and certification targets. We map them to the right compute and connectivity.
  2. 02Prototype on a kitStart with an evaluation kit and validate perception, interfaces and software on the target compute architecture.
  3. 03Integrate the production systemIntegrate compute, carrier, sensors and connectivity, with reviews for EMC, thermal performance and the documentation required by the programme.
  4. 04Manufacture and maintainMove to series production with lifecycle management, telemetry and staged updates through Clea.

Frequently asked

Tell us what your robotic system needs to sense, decide and control. We will recommend the compute, connectivity and lifecycle strategy.