If you're just getting started with ROS or want to build your own ROS robot, choosing the right control board is very important.
Should you choose the powerful NVIDIA Jetson development board or the classic, classic Raspberry Pi ROS solution?
Different ROS development boards can determine your project's potential, development experience, and ultimately its success.
This guide will analyze the features of mainstream ROS robot control boards to help you make the most informed choice based on your project needs.
Step 1: Define Your ROS Project Requirements
General users will consider the following three aspects:
1. Computing Power Requirements:
Does my robot need to perform computationally intensive tasks like real-time image recognition and SLAM mapping? Or does it only require basic motion control and sensor data fusion?
2. Power Consumption and Size:
Does the robot have strict size, weight, and endurance requirements? (For example, a legged robot or a small drone)
3. Budget and Ecosystem:
What's my budget?
Do I need extensive community support and easily accessible accessories?
Step 2: Comparison and Evaluation of Two Mainstream ROS Control Board
The following is an introduction to several common ROS controllers currently on the market.
1. NVIDIA Jetson Series: The King of AI Computing Power
Representative Products: NVIDIA Jetson Orin AGX, NVIDIA Jetson Orin NX, NVIDIA Jetson Orin NANO, NVIDIA Jetson Nano B01
Advantages:
Superior AI Computing Power: The built-in GPU core delivers AI performance ranging from several TOPS to hundreds of TOPS, enabling real-time processing of multi-channel camera data and easily running TensorRT-accelerated deep learning models.
Perfect Support for Different ROS Versions: The official JetPack SDK includes libraries such as CUDA, cuDNN, and TensorRT, seamlessly integrating with ROS and ROS 2 vision and AI packages.
Abundant Onboard Interfaces: Numerous CSI, PCIe, and USB ports facilitate connection to various sensors, such as lidar, depth cameras, and voice recognition modules.
Disadvantages:
High price: Relatively high power consumption, resulting in high temperatures over extended periods of high load; it is recommended to use with efficient cooling equipment.
Below is a performance comparison chart of several Jetson development boards.
Best suited for:
A: For ROS beginners, students, and educational projects, the Jetson NANO Developer Kit is recommended.
Jetson NANO B01 Developer Kit Shipping List
From basic system setup and GPIO hardware control to AI vision applications, and from ROS1/ROS basics to advanced applications, extensive tutorials and open-source code are provided for your reference.
Get a complete development experience with minimal investment, making it ideal for budget-conscious users seeking a solid foundation.
B: For those looking to build smart cars and deploy common AI models, the Jetson ORIN NANO development board is recommended.
Capable of processing multi-channel sensor data in real time and running complex deep learning models (such as YOLO object detection and SLAM), the Jetson ORIN NANO offers a significant performance boost, giving your smart car a true "smart brain."
Delivering 34 to 67 TOPS of AI performance, several times the performance of the Jetson NANO, it can process high-resolution video streams in real time and run larger, more accurate neural network models. It can easily handle advanced tasks such as multi-camera visual SLAM, 3D perception, speech recognition, and autonomous driving, giving your smart car the perception and decision-making capabilities it needs.
Note:
The Jetson ORIN NANO development board is typically a single-board device. You may need to configure your own power supply, storage (NVMe SSD or SD card), cooling solution, and carrier board, which requires additional investment and hardware knowledge.
If you want to quickly configure all external accessories, you can directly choose these developer kits and the Jetson ORIN NANO ROS Robot Car Kit.
C: Engineers and teams interested in building complex robots, biomimetic robots, deploying large AI models, or engaging in cutting-edge research and product development should consider the Jetson ORIN NX board.
It offers up to 117/157 TOPS of computing performance, far exceeding the ORIN Nano and NANO series. This enables it to process massive amounts of data from multiple sensors, such as LiDAR, multi-cameras, and IMUs, in real time, and to locally deploy and run tailored large AI models (such as LLMs and multimodal models), empowering robots with enhanced environmental understanding, autonomous decision-making, and human-machine interaction capabilities.
Note:
The Jetson ORIN NX development board generally requires deeper hardware integration capabilities. You'll need a carrier board, a power solution that meets power requirements, and an efficient active cooling system. This provides developers with extreme flexibility, but also presents challenges.
We recommend choosing a ROS robotics kit based on the Jetson ORIN NX and completing your own development based on detailed tutorials and open-source code.
2. Raspberry Pi Series: Complete Development Ecosystem and Community Environment
Representative Products: Raspberry Pi 5 (4GB/8GB/16GB)
Advantages:
Mature Community: With the largest user base and community support, you can find answers to almost any question you may have. Detailed installation and configuration tutorials are available for both ROS1 and ROS2.
Excellent Value: Get a fully functional Linux computer at a very low cost, offering an excellent value for money.
Abundant Accessories: A comprehensive hardware ecosystem includes a variety of HAT expansion boards, heat sinks, enclosures, and cameras.
Disadvantages:
Limited AI computing power, especially inability to process video and images.
Best suited for:
ROS beginners, educational projects, low-to-medium-speed mobile robots, service robots with low AI requirements, and projects requiring rapid prototyping.
The ROS robot kits based on Raspberry Pi shown below are also a good choice for you.
3. RDK (D-robotics Robot Development Kit): A new embedded development board offering cost-effectiveness and ease of use.
Representative Products: RDK X3 Board, RDK X5 Board
Advantages:
On board 8-core A55 CPU, and comes with almost the same interface configuration as Raspberry Pi. It provides up to 10TOPS equivalent computing power, supports Ubuntu22.04 system, and can efficiently run cutting-edge AI models such as Transformer, RWKV, Occupancy Networks, Stereo Vision, Perception, etc., to meet complex computing needs such as robot navigation, environmental perception, and multimodal interaction, so that your projects can be quickly implemented.
D-robotics provides it with 100+ open source algorithms and applications to help customers develop their own projects efficiently.
Disadvantages:
Relatively low community activity: Compared to Jetson and Raspberry Pi, which have global communities, its international influence and community size are still growing.
Brand recognition: As a relatively new brand, it lacks visibility.
Best suited for:
ROS beginners and educational institutions: Ideal for robotics teaching and training, allowing students to quickly focus on ROS algorithms rather than environment configuration.
Developers seeking a cost-effective Jetson alternative: Individuals or teams seeking Jetson-level AI computing power while reducing hardware costs and initial configuration time.
Conclusion and Final Recommendations
For developers seeking superior AI performance and with ample budgets:
If your project centers around computer vision, deep learning, or autonomous navigation, requiring substantial edge AI computing power but is constrained by the price of high-end products, the NVIDIA Jetson series (such as the Jetson Orin Nano/NX) is the ideal solution.
For ROS enthusiasts new to development who need the support of a comprehensive ecosystem and community resources:
If you're just beginning your ROS development journey and your top priority is finding a platform with strong community support, numerous open source projects, and comprehensive learning resources, the Raspberry Pi 5 is undoubtedly your safest and smartest choice. Not only does it run ROS1 and ROS2 flawlessly, its low barrier to entry and vast ecosystem ensure that most of your questions have already been answered.