The Alius DCU (Domain Control Unit) is a high-performance edge computing and control platform designed for next-generation intelligent robotic systems. Built on the Rockchip RK3588S SoC, it achieves a seamless integration of real-time control and high-performance heterogeneous computing on a single board.

Product Positioning

The DCU Controller fills the gap between traditional PLCs and general-purpose industrial PCs. It combines the determinism and reliability of industrial-grade real-time control with the high-performance computing capabilities required by modern AI applications. This unique positioning makes it an ideal main control unit for complex systems such as humanoid robots, autonomous mobile robots (AMRs), and industrial robot arms.

Core Architecture

The DCU adopts a heterogeneous computing architecture that integrates multiple computing units on a single SoC:

  • CPU Subsystem: 4× Cortex-A76 + 4× Cortex-A55 octa-core configuration. The A76 cores run at 2.4 GHz and handle high-performance computing tasks, while the A55 cores run at 1.8 GHz and focus on real-time control tasks, achieving physical isolation between computing and control.
  • GPU Subsystem: Mali-G610 MP4, supporting OpenGL ES 3.2, Vulkan 1.2, and OpenCL 2.2, suitable for vision processing, UI rendering, and lightweight inference acceleration.
  • NPU Subsystem: 6 TOPS neural network acceleration unit, supporting mixed-precision computing with INT4/INT8/INT16/FP16, capable of running models from TensorFlow, PyTorch, ONNX, and other frameworks.
  • Real-time Coprocessor: Independent RTOS core that guarantees determinism for control cycles through hardware interrupts, supporting hard real-time tasks such as EtherCAT master station and CAN-FD bus scheduling.

Software Platform

The DCU Controller supports multiple operating system deployment options:

  • Linux User Space: Supports mainstream distributions such as Debian, Ubuntu, and Buildroot, providing a complete user-space ecosystem including Python, C/C++, ROS 2, Docker, and other development toolchains.
  • RTOS Coprocessing: The integrated real-time operating system handles time-critical tasks and communicates efficiently with Linux through shared memory and interrupt mechanisms, ensuring microsecond-level determinism for control cycles.
  • Heterogeneous Scheduling Framework: Provides a unified task scheduling API, allowing developers to transparently allocate computing tasks to appropriate processing units without worrying about underlying hardware details.

Interfaces and Connectivity

The DCU provides a rich set of industrial-grade interfaces to meet the diverse connectivity requirements of robotic systems:

  • 2× CAN-FD: Supports CAN-FD communication up to 5 Mbps, with electrically isolated design that can directly connect to joint modules, sensors, and actuators.
  • 2× RS-485: Isolated RS-485 interfaces supporting industrial protocols such as Modbus RTU, suitable for connecting legacy industrial equipment.
  • 2× EtherCAT: Supports EtherCAT master mode, enabling high-performance real-time control networks to connect multiple slave devices.
  • Gigabit Ethernet: For high-speed data transfer, remote debugging, and system integration.
  • USB 3.0: Can be used to connect high-speed peripherals such as cameras and LiDAR.
  • MIPI-CSI×2: Supports connecting two MIPI cameras for visual perception and SLAM.
  • MIPI-DSI: Can be used to connect display panels for local HMI functionality.

Power and Reliability

The DCU Controller is designed for industrial environments:

  • Wide Voltage Input: 9-36 VDC input range, adaptable to various power configurations in robotic systems, supporting mainstream 12V and 24V industrial power supplies.
  • Electrical Isolation: CAN and RS-485 interfaces adopt electrically isolated design with isolation voltage up to 2500 Vrms, effectively preventing ground loops and electrical interference.
  • Industrial-grade Components: All critical components are selected from industrial-grade temperature range (-40°C to +85°C) products, ensuring long-term stable operation in harsh environments.

Application Scenarios

The DCU Controller has been validated in multiple robotic application scenarios:

  • Humanoid Robot Control: Serving as the central controller, responsible for motion planning, attitude control, environmental perception, and human-robot interaction with multi-task parallel processing.
  • Autonomous Mobile Robots (AMR): Serving as the core of navigation and control systems, running SLAM, path planning, and obstacle avoidance algorithms, while controlling chassis motors through EtherCAT.
  • Industrial Robot Arms: Serving as the controller, running kinematic calculations, trajectory planning, and force control algorithms, supporting both EtherCAT and CAN-FD industrial buses.
  • Composite Robotic Systems: Simultaneously controlling mobile chassis and robotic arms to perform complex operation tasks such as liquid handling, assembly, and inspection.