To not miss my updates, remember to click the top right corner – view public account – set as a star, and send me a star.
Recently, ChatGPT has created a whirlwind, greatly increasing everyone’s enthusiasm for artificial intelligence. More and more applications are starting to integrate AI, such as AIoT.
To enable hardware engineers to quickly get started with AI projects, Darwin has teamed up with Dajie Electronics to launch a series of videos on AI edge computing. The four-part video series uses three AI development boards: LattePanda 3 Delta, BeagleBone AI64, and ODYSSEY-X86J4125-V2 to guide you step by step in creating AI projects from scratch.
#
Why choose these three boards for AI edge computing
There are many AI development boards on the market, and choosing the right AI development board, running demos, and getting hands-on experience can quickly lead to project development.
The three development boards we selected each have their strengths:
1) Application Scenarios: Different application scenarios require different hardware and software support. These three development boards have relatively broad application scenarios and can be used in almost all AI applications, including IoT, industry, and automotive, etc. For example, for autonomous driving applications, you can use the BeagleBone AI64, which can be used in many ADAS L2 level autonomous driving solutions. As for LattePanda and Odyssey, in addition to regular AI applications, when connected to a screen and mouse and keyboard, they become a mini PC.
2) Processor Type: Usually, AI development boards need to be equipped with high-performance processors, such as Intel, NVIDIA, Qualcomm processors, and many use domestic processors. Among these three, LattePanda and Odyssey use Intel processors, with LattePanda using the Celeron N5105 quad-core processor with a maximum boost frequency of 2.9GHz, while Odyssey uses the Celeron J4125, which is also a quad-core processor. The processor of BeagleBone AI64 is TI’s TDA4VM, which has two 64-bit Arm® Cortex®-A72 cores with a frequency of up to 2.0GHz, and also has an image signal processor (ISP) and multiple vision processing accelerators (VPAC). Basically, all AI development boards emphasize strong performance.
3) Memory and Storage: AI development boards need to deploy artificial intelligence algorithms and require large-capacity memory and storage space to support high-performance AI applications. These three boards have relatively large storage space, with LattePanda and Odyssey both having 8G LPDDR4 RAM and 64G eMMC storage, and supporting NVME solid-state drives, allowing the system and applications to be installed on the solid-state drive. BeagleBone’s storage space is slightly smaller, with 4G LPDDR4 RAM and 16G eMMC storage, but it is basically sufficient.
4) Hardware Resources: These three AI development boards are equipped with rich resources, including gigabit Ethernet, HDMI 2.0, USB 3.0, WIFI 6, Bluetooth 5.0, various sensors, etc. They also have certain expandability, such as Odyssey, which integrates an Arduino co-processor and has pins compatible with Raspberry Pi 40pin, while BeagleBone AI64 is compatible with Beagle series expansion boards.
5) Software Support: It is essential to provide a wealth of development tools, libraries, and sample codes. These three boards do not require specific software packages when performing ROS development or edge controller usage, and they have strong support. Programs developed on personal PCs can be directly deployed without excessive modifications.
#
Preview of AI Projects
Based on these three development boards, we have invited four senior engineers who are proficient in AI development boards to provide hands-on teaching.
1) LattePanda: We invited senior maker Drift Brother and Hao Nan to share their experience using LattePanda to create an open-source donkey car based on machine vision and neural networks for automated driving. Although it is an RC car, it involves multiple aspects of the technology stack, from the hardware structure of the car to the software level. They will teach everyone step by step how to install and configure the Linux operating system when you get a new LattePanda, including network setup, package management, Python virtual environment, OpenCV visual framework, TensorFlow neural network framework, Keras frontend, RC remote control debugging, cloud GPU training, and automation processing.
2) ODYSSEY-X86J4125-V2: We invited senior engineer Industrial Control Brother to teach how to use the Odyssey development board to develop a smart IoT server for the whole house, deploying the IoT platform on the development board, covering various intelligent operations, including using Odyssey’s powerful AI computing power for facial recognition and electronic cat-eye detection.
3) BeagleBone AI64: Compared to the other two development boards, BeagleBone AI64 is the latest release, and the documentation is relatively scarce, which may pose a challenge for beginners. Therefore, senior AI engineer Xiao He will guide everyone step by step in deploying the environment, demonstrating typical AI application models, and displaying results on the screen while generating performance data to help everyone quickly deploy applications.
These three courses are all hands-on teaching, starting from how to deploy the system, update the code, run the official demo, and finally use these boards to complete an actual project.
Scan the QR code to follow Dajie Electronics on Bilibili
Next, Darwin will provide detailed updates on each project’s introduction, stay tuned.
Project Sharing | Electrical Competition Series | Artificial Intelligence | Postgraduate Entrance Examination
Must-Know Knowledge Points | Graduation Design | Switch Power Supply | Job Hunting
We are Nimo, the founder of Darwin, a sister who only talks about technology and does not flirt. Darwin is an online education platform aimed at serving professionals in the electronics industry, providing training videos on skills covering popular topics in various niche fields such as embedded systems, FPGA, and artificial intelligence. We tailor hierarchical learning content for different groups, such as common knowledge points, disassembly assessments, electrical competitions/smart cars/postgraduate entrance examinations, etc. Welcome to follow us.
