Contec Americas Blog

How do I Know if Raspberry PI, NVIDIA Jetson Nano, or Intel NUC are Appropriate Computers for my Industrial Robotic Application?

Written by Sample HubSpot User | Aug 2, 2021 8:30:56 PM

Regularly engineers look at three options when looking to source Single Board Computers for Robotic Applications: Raspberry Pi, NVIDIA Jetson Nano, and Intel NUC. But how you should choose between any of this or a customized solution? We will provide a guide in this blog post regarding the advantages of each option for Autonomous Mobile Robots (AMRs) or Automatic Guided Vehicle (AGVs).

Let’s assume for a moment that you are a golf enthusiast looking to acquire golf clubs. You need to considerer several factors before purchasing them: body type, height, previous experience and education, shaft flexibility and length, variety of clubheads, most frequented golf fields, and your swing, among others. In other words, brand is not as important as how often, where and who will use the golf clubs. This applies to SBCs as well.

When you develop and design a robot, you never start researching SBCs based on brands. Or at least, you should not. You must start at the end: with the user and the application requirements. You can read more about robotics unique hardware requirements in this blog post. You should pay special attention to the following factors when deciding the SBC for your robotic application:

  • Processing capabilities: processor and chipset
  • Memory: How much RAM and ROM or HD the application requires
  • Footprint: The physical size and weight of the single board computer
  • I/O: How many I/O ports the application needs and why

If you want to read more on the factors that affect what computing capabilities to choose for a mobile robotic application, you can read this blog.

Nevertheless, if you still need to know if Raspberry Pi, NVIDIA Jetson Nano, or Intel NUC computers are an option for your business, we will go over some important details you need to consider.

Raspberry Pi

This is an Open-Source project that includes both software and hardware. Its widespread use started thanks to its educational and maker focus. In fact, most Raspberry PI boards are appropriate for school and university robotics projects. The community is a deciding factor when people choose these boards because you can leverage the experience from makers, students, and institutions all over the world to troubleshoot or borrow designs/programming. Thus, when engineers start working, they look for the boards they are familiar with.

Given the stringent requirements in industrial environments in terms of temperature, size, water or dust, Raspberry Pi launched a board intended for commercial applications for those engineers who were trying to apply their previous experience with these boards in their new working environment. In addition, you can add rugged features to a Raspberry Pi 4 to withstand harsh industrial environments. Learn more here.

The main selling point for industrial Raspberry PI boards is their footprint (credit card size) and their price. You need to be aware that although software for research and maker purposes might be free, for commercial applications you may need to pay for licenses or be able to share code (which might be incompatible due to commercial intellectual property rights).

Finally, the worldwide community of makers might be able to help you sort some of the commercial challenges, but they might not be able to tackle them all. As a result, this might not be the same advantage you get as a student.

NVIDIA Jetson Nano

NVIDIA comes from a different background. They basically created the GPU back in 1999. Their graphic boards are designed to handle gaming, designing, and professional graphics. Consequently, NVIDIA Jetson Nano boards are better equipped to handle data intensive robotic applications related to Deep Learning, Machine Learning, Pattern Recognition, and Artificial Intelligence. This is possible thanks to the GPU parallel processing ability.

In this case, you might not rely on the same large worldwide community of makers as Raspberry Pi but you do have the support from an innovative, robust company who has been designing SBCs for commercial applications since the past century. NVIDIA’s offering includes some software which is already tested with the boards. This can be helpful in terms of compatibility and cost.

One of the NVIDIA Jetson Nano boards has a 128-core Maxwell GPU at 921 MHz which is more powerful than the Raspberry PI 4 GPU. You can see the Jetson Nano offering here.

Intel NUC

Intel’s response for the SBCs market both for home and commercial applications is the NUC. These computers can be acquired as boards or complete with chassis depending on the intended use. The selling point is the wide range of processors: Pentium, Celeron, Core i3 or Core i5. As a result, you will choose the processor that can handle the application workload. As a matter of fact, the latest processors work smoothly in data intensive applications, like Artificial Intelligence. Thus, Intel suggests using NUC for immersive gaming, digital signage, remote meetings, and vivid home entertainment.

Size is comparable to the Jetson Nano and adjusts to the limited footprint of robotic applications. However, in terms of price, these SBCs can be more expensive than the first two options in this blog.

Moreover, Intel NUC is compatible with Windows 10 Pro. If you are planning to work with Windows this is the platform that will provide the best compatibility as Raspberry Pi and NVIDIA will work best with Linux.

In AMRs/AGVs power consumption is a vital factor. If the robot has a charging station close by, huge batteries or it is connected directly to the power source, the designer can choose the most powerful processor and guarantee a smooth operation. However, AMRs/AGVs typically have limited power sources which forces developers to choose less powerful processors and perform some analytics in a remote brain (server). From the three options we are analyzing, Raspberry PI and Jetson Nano require less power than Intel NUC.

The last interesting feature of Intel NUC is that it allows dual and triple monitor connectivity. Raspberry Pi only allows dual monitor and NVIDIA Jetson Nano, just one. You can check out Intel NUCs offering here.

Summary of Tech Specs

Features

Raspberry PI 4 Model B

Jetson Nano Developer Kit

Intel® NUC 11 Pro Board NUC11TNBv7

Processor

Broadcom BCM2711, Quad core Cortex-A72 (ARM v8) 64-bit SoC @ 1.5GHz

CPU: Quad-core ARM A57 @ 1.43 GHz

GPU: 128-core Maxwell

Intel® NUC Board with 11th Generation Intel® Core™ Processors

Memory

2GB, 4GB or 8GB LPDDR4-3200 SDRA

microSD (not included)

4 GB 64-bit LPDDR4 25.6 GB/s

microSD (not included)

64 GB, DDR4-3200 1.2V SO-DIMMs

Footprint

85mm x 56mm

(3.3” x 2.2”)

69mm x 45mm

(2.7” x 1.77”)

101mm x 101mm

UCFF (4" x 4")

I/O

2.4 GHz and 5.0 GHz IEEE 802.11ac wireless, Bluetooth 5.0, BLE

Gigabit Ethernet

2 USB 3.0 ports; 2 USB 2.0 ports.

2 × micro-HDMI ports (compatible with Dual displays)

Micro-SD card slot for loading operating system and data storage

2x MIPI CSI-2 DPHY lanes

Gigabit Ethernet, M.2 Key E

HDMI and display port

4x USB 3.0, USB 2.0 Micro-B

PCI Express, 4 USB Ports, 1 SATA Port, Integrated LAN, 1x Thunderbolt™ 4, 1x Thunderbolt™ 3

Compatible with dual and triple displays.

Operating System

Raspberry Pi OS (previously called Raspbian) is the recommended operating system

Linux

NVIDIA JetPack SDK

 

Windows 10, 64-bit*, Windows 10 IoT Enterprise*, Red Hat Linux*, Ubuntu 20.04 LTS*

Power Consumption at 100% usage

6 Watts

10 Watts

49 Watts

Price

From $35

From $59

From $877

 

Customized Solution

Finally, the feature that is missing in the past three options is Industrial Certifications and Protections. When an AMR/AGV is intended to work at factories, mines, or medical applications, certifications are a crucial factor. An experienced Hardware OEM can guide you through the process and offer an SBC that can adapt to the stringent environment.

Also, a Hardware OEM can make sure all connectors included in the customized SBC can withstand vibrations and avoid unintended disconnections. Plus, they can match the specific application requirements to chipsets, firmware, processor, memory, and footprint. If you have several products that require the same capabilities, this is an attractive advantage. Therefore, you could apply one SBC design to all your products and save on manufacturing and repairing costs. This is particularly helpful with the worldwide shortage of electronic components we are experiencing nowadays. You can read a recent success story here.

As you can see, choosing the SBC for your AMR/AGV depends on the application requirements and expected user experience. The decision should be made based on size, processing capabilities and required level of protection/certification. There are several good off-the-shelf options in the market that are helpful during the first stages of design and prototyping. Once your AMR/AGV is ready for mass production you might need to leverage an experienced Hardware OEM to develop a customized board solution.

You can read more about hardware unique requirements in our Robotics Blog Series or subscribe to receive more information about this or similar topics.