D-Robotics RDK X5 Development Kit Sunrise 5 AI Chip Cortex A55 10TOPs

TheD-Robotics RDK X5is an intelligent robot development board powered by theSunrise 5 AI SoC, delivering10 TOPSof neural processing power through its dedicated BPU (Brain Processing Unit). Built around anocta-core ARM Cortex-A55processor, the RDK X5 provides a credit-card-sized (85 56 mm) platform with exceptional AI inference capabilities, rich peripheral interfaces, and seamless integration with the TogetheROS.Bot framework making it an ideal choice for robotic perception, edge AI deployment, and autonomous navigation applications.

Whether you are building an autonomous mobile robot, a vision-guided manipulator, or an edge AI prototype, the RDK X5 combines high-performance computing with a beginner-friendly software ecosystem. The board supports out-of-the-box deployment of advanced algorithms includingYOLO object detection, stereo vision, occupancy networks, and even large language models (LLMs)at the edge. WithWi-Fi 6, Bluetooth 5.4, dual MIPI CSI cameras, HDMI output, PoE Ethernet, CAN FD, and a 40-pin GPIO header, the RDK X5 covers virtually every connectivity requirement for modern robotic systems.

Available in4GBand8GB LPDDR4xmemory variants, the RDK X5 is offered in bothSingle BoardandBasic Kitconfigurations to suit different project stages from bare-metal prototyping to plug-and-play development.

Features

• Sunrise 5 AI SoC with 10 TOPS BPU Dedicated Brain Processing Unit for high-efficiency neural network inference, supporting INT8/INT4 quantization and advanced models including Transformer, RWKV, and CLIP at the edge
• Octa-Core Cortex-A55 @ up to 1.8 GHz Powerful application processor capable of handling multi-task workloads, real-time control loops, and concurrent AI pipelines
• 4GB / 8GB LPDDR4x Memory Options 4GB variant ideal for standard YOLO inference and robotics control; 8GB variant optimized for large edge models (LLM), multi-camera stereo vision, and complex perception stacks
• Dual 4-Lane MIPI CSI-2 (MIPI v2.1) Connect up to two high-resolution cameras simultaneously for stereo depth estimation, multi-view perception, or simultaneous RGB + depth sensing
• HDMI Output + 4-Lane MIPI DSI 1080P@60fps display via HDMI and an additional MIPI DSI interface for embedded display panels ideal for HMI and debug visualization
• 4 USB 3.0 Host + USB 2.0 Type-C OTG High-bandwidth USB 3.0 ports for cameras, LiDAR, and external storage; USB 2.0 Type-C for device mode and serial console
• Gigabit Ethernet with PoE Support 1000Mbps RJ45 port with Power-over-Ethernet capability for simplified single-cable deployment in robotic and industrial environments
• Wi-Fi 6 (2.4/5 GHz Dual-Band) + Bluetooth 5.4 Wireless connectivity with onboard high-performance antenna and IPEX connector for external antenna options
• CAN FD + 40-Pin GPIO Header 28 GPIOs with native SPI, I2C, I2S, PWM, and UART support; CAN FD bus for motor controllers and industrial communication
• Onboard 3.5mm Audio Jack Combined audio input/output for voice interaction, sound localization, or TTS-based HMI feedback
• RDK OS (Linux-Based) + RDK Studio IDE Complete development environment with RDK Studio for application development, RDK Nodehub for one-click app deployment, and RDK Cloud integration
• TogetheROS.Bot (ROS 2 Compatible) Native integration with the TogetheROS.Bot middleware stack, providing ROS 2 node support, Livox LiDAR drivers, and a rich open-source model zoo (rdk_model_zoo) with 296+ GitHub stars
• Rich AI Algorithm Library Pre-integrated open-source algorithms including YOLO, Occupancy Networks, Stereo Vision, Perception, and World Model all deployable with one-click tooling
• Edge LLM Support Deploy large language models at the edge with native Volcengine LLM Gateway integration (complimentary traffic included), enabling on-device conversational AI
• Compact Form Factor: 85 56 20 mm Credit-card-sized footprint with versatile mounting options; only 5V/5A power input via Type-C PD
• Wide Operating Temperature: -20C ~ 60C Suitable for indoor lab environments as well as outdoor robotic deployments

Specifications

Camera & Display

Connectivity

USB Ports

Expansion & I/O

Power & Physical

Variant Comparison

Reference Documents:

• D-Robotics Developer Community Documentation & SDK Downloads
• RDK OS Image Build Tools (GitHub)
• RDK Model Zoo Pre-trained AI Models for X5 (GitHub)
• RDK X5 Product Page (Waveshare)

Application

• Autonomous Mobile Robots (AMR) Leverage dual MIPI CSI cameras for visual SLAM, obstacle avoidance, and path planning with the 10 TOPS NPU accelerating real-time perception pipelines
• Robotic Vision & Perception Deploy YOLO object detection, stereo depth estimation, and occupancy networks for robotic manipulation, pick-and-place, and scene understanding
• Edge AI & LLM Deployment Run large language models (Transformer, RWKV) directly on-device with 4GB/8GB memory, enabling conversational AI, intelligent decision-making, and voice interaction without cloud dependency
• ROS 2 Robot Development Native TogetheROS.Bot integration provides ROS 2 nodes, Livox LiDAR drivers, and sensor fusion frameworks for rapid robot prototyping
• Smart Surveillance & Security Multi-camera real-time video analytics with AI-powered detection, classification, and tracking at the edge
• Education & Research Labs Ideal platform for universities and research labs exploring AI, computer vision, robotics, and embedded machine learning
• Industrial Inspection CAN FD connectivity for industrial automation, combined with high-performance vision processing for quality control and defect detection
• Drone & UAV Payload Compact 85 56 mm form factor with PoE Ethernet and lightweight design suitable for aerial AI computing platforms

Packing List

Single Board

Basic Kit

FAQ

Q1: What is the difference between the 4GB and 8GB variants?
The only hardware difference is the RAM capacity:4GB LPDDR4xvs8GB LPDDR4x. All other specifications CPU, NPU (10 TOPS), interfaces, and physical dimensions are identical. Choose4GBfor standard YOLO inference, basic robot control, and Linux development. Choose8GBfor running larger edge models (LLM, Transformer), multi-camera stereo vision, or memory-intensive AI workloads.

Q2: What is the difference between Single Board and Basic Kit?
TheSingle Boardincludes only the bare RDK X5 development board suitable if you already have compatible accessories (power adapter, TF card, heatsink). TheBasic Kitbundles the board with a64GB TF card (pre-loaded with RDK OS), heatsink, SD card reader, USB Type-A to Type-C cable, and a 27W PD power adapter with multi-region plug adapters everything needed to power on and start developing immediately.

Q3: Can I run ROS 2 on the RDK X5?
Yes. The RDK X5 natively supportsTogetheROS.Bot, D-Robotics’ ROS 2-compatible middleware stack. It provides standard ROS 2 nodes, sensor drivers (including Livox LiDAR vialivox_ros_driver2), and navigation/SLAM packages. Therobot_dev_configrepository provides the build entry point for TogetheROS.Bot on X5.

Q4: What AI models and frameworks are supported?
The RDK X5 supports a wide range of AI models through theRDK Model Zoo(GitHub), includingYOLO (object detection), Occupancy Networks, Stereo Vision, Perception, and World Modelalgorithms. Edge LLM deployment is supported with frameworks like Transformer and RWKV. All models benefit from the 10 TOPS BPU with one-click deployment tooling via RDK Studio.

Q5: How do I power the RDK X5?
The board requires5V/5A (25W)power input via the USB Type-C port with PD (Power Delivery) protocol. The Basic Kit includes a 27W PD power adapter. The Single Board variant requires you to provide a compatible USB-C PD power supply.PoE (Power over Ethernet)is also supported through the Gigabit Ethernet port, enabling single-cable power + data connection.

Q6: What cameras are compatible with the MIPI CSI interfaces?
The RDK X5 provides2 4-lane MIPI CSI-2 (MIPI v2.1)camera interfaces. D-Robotics offers official RDK stereo camera modules. Additionally, third-party MIPI cameras compatible with thehobot_mipi_camdriver can be used. The dual-camera setup supports stereo depth estimation for 3D perception and SLAM applications.

Q7: Is the 40-pin GPIO header compatible with Raspberry Pi HATs?
The RDK X5 features a40-pin GPIO headerwith a similar physical layout to Raspberry Pi, providing 28 GPIOs at 3.3V logic level with native support forSPI, I2C, I2S, PWM, and UART. While the pin layout may be similar, electrical compatibility with specific Raspberry Pi HATs should be verified on a case-by-case basis due to potential voltage and signal differences.

Q8: What operating system does the RDK X5 run?
The RDK X5 runsRDK OS, a Linux-based operating system maintained by D-Robotics. System images can be built from source using thex5-rdk-gentools on Ubuntu 18.04/20.04/22.04. Development is facilitated throughRDK Studio(IDE) andRDK Nodehub(one-click app launcher). Cloud integration is available via RDK Cloud.

Q9: Can the RDK X5 be used for LiDAR-based SLAM?
Yes. With4 USB 3.0 ports, Gigabit Ethernet, and CAN FD, the RDK X5 can connect to various LiDAR sensors including Livox, RoboSense, and RPLiDAR via USB or Ethernet. D-Robotics provides an officialLivox ROS 2 driver, and the TogetheROS.Bot stack includes navigation and SLAM packages. The 10 TOPS NPU can simultaneously handle perception, mapping, and planning tasks.

Q10: What is the warranty policy for the RDK X5?
Please refer to the manufacturer’s warranty policy as specified on the product packaging or official documentation. For warranty claims and technical support, contact the point of purchase or visit theD-Robotics Developer Community.