Multi-FDCAN integrated controller in development — one board, from industrial arms to humanoids

An R&D update — SaeA-ICT is currently developing a multi-FDCAN integrated controller board. It is not yet a productised SKU; host integration, communications, and the web GUI are validated, and the project is now moving into motor-driver integration and the production-PCB phase.

One-line summary

A multi-FDCAN integrated controller that mounts next to an NVIDIA Jetson series or a Raspberry Pi host and orchestrates many motor groups in parallel. The same board is being designed to cover everything from industrial arms to a full humanoid.

Key features

• Multi-channel FDCAN backbone — motor groups are split across several FDCAN channels, so traffic on one channel does not bleed into the others. Even on a many-axis system the response stays predictable.

• Built-in 3-Port 100Mbps Ethernet switch — host, controller, and a debug PC sit on the same single cable port, simplifying field wiring and pairing naturally with ROS2-style network applications.

• Direct connection to NVIDIA Jetson and Raspberry Pi — designed to mate with the standard host header so a Python / C++ / ROS2 application on the host can drive motor groups directly, with no separate gateway.

• Multiple host-link paths — the main host link is paired with a secondary path so a single blocked route does not stop the work.

• Dedicated emergency-stop line — a separate signal line connects host and controller outside the message queue, guaranteeing the fastest possible stop in any state.

• Visible runtime state — an on-board status display plus a host-side web GUI, so operation, faults, and link health show up immediately on a single cable.

Coverage — one board, a wide robot spectrum

Thanks to the multi-channel FDCAN structure, the same controller covers:

• Industrial robot arms / multi-axis manipulators — 6–7 axis arms, desktop cobots, and other lower-axis-count platforms.

• Modular multi-joint systems — multi-joint platforms with grippers, hands, or tool changers.

• Full humanoid bodies — both arms, both legs, the waist, the head, and the fingers all distributed across the channels of a single controller.

The core design intent: "more axes does not mean more controllers".

Why this board exists

In robotics, the layer between the host "brain" (Jetson / Raspberry Pi) and the motor drivers is normally stitched together from several small controllers, and the wiring complexity, the cable count, the consistency of emergency-stop, and the debug environment have to be re-invented for every project. This board standardises that mid-layer into one card, so every in-house robotics project can start from the same interface.

Status (May 2026)

• ✅ Direct host (Raspberry Pi) communications — both the primary and the secondary path verified end-to-end.

• ✅ Multi-channel FDCAN sanity validated (loopback level).

• ✅ Host-side integrated web GUI — board state, pin map, and live commanding are visible from a browser.

• ✅ A status-monitoring daemon that comes up automatically when the host boots.

• 🔄 Motor-driver group integration — planned next.

• 🔄 Production-PCB port + EMC / EMI evaluation — planned next.

What is next

A full-body demo combining the board with our in-house BLDC driver assets, then a first production PCB for partner / customer evaluation, with the goal of having this board sit at the centre of our humanoid and cobot product lines. Reach out if you would like to discuss applying it to your platform.

Do you have an idea?

From device design to firmware, content integration, and mass production — SaeA-ICT is with you.