Prerequisites for using large models (must read)

Example cases related to large models, important prerequisite: you need to connect to a wifi that can access the Internet. For specific connection methods, you can go to [1. Basic operations of Raspberry Pi system] --> [6. Connect to WiFi] to see the specific operations. This tutorial will briefly explain it.

1. First download the xgo Bluetooth remote control APP.
2. Enter the app, click the network configuration information in the upper right corner, fill in the wifi name and password, and click to generate a QR code.
3. The robot dog selects the wireless networking function through the buttons on the screen, and points the QR code generated by xgo at the robot dog's camera.
4. After waiting for a while, the wireless networking function will automatically exit, indicating that the wifi is connected

This tutorial is about the configuration of AI large models. If you want to run the large model related cases of the boot program example gameplay normally, you need to follow the steps below to configure it.

1. Overseas version of OpenRouter

Some online large models of the overseas version of the boot example use the middleware of the OpenRouter platform, which is a third-party platform. This large model can be called for free 50 times a day. If you want to continue using it, you can refer to the usage rules for expansion.

The following cases require the use of this large model: Q&A, Embodied intelligence, Smart dog, and Picture to text

OpenRouter operation steps

1. First log in to the following official website

https://openrouter.ai/

2. Then log in. You can use multiple ways to log in. This article recommends using a github account to log in to this platform.

3. Enter the API-KEY page and create one

4. Then copy the created key and save it on your computer. You can only copy this key when you create it for the first time. You cannot view it again after leaving this interface.

5. Then we find a free model to test whether the registered account can be used normally

6. If it is a normal reply, it means it can be used normally.

7. The created API-KEY can directly use the free model. If you want to use the non-free model, you need to purchase the non-free model with your credit card. There is no tutorial or explanation, so you can study it yourself.

2. Overseas version of Qwen

Some online large models in the overseas version startup example use the visual large model of Qwen. This platform is a third-party platform, and the large model is a powerful visual model provided by Alibaba. After the registration and authentication is completed, some large models can be used for free, and more models can be used according to this website.

The following cases require the use of this large model: Control object tracking

1. Open the website and register

Register account

2. After registration, enter the contact information according to your country and get certified. Without certification, you cannot get the free large model

3. Enter the large model, select Qwen-VL-Max, and get it for free Take part of the large model quota

4. Register your own API Key

5.Register your own OpenAI Compatible Endpoint

Then remotely access the robot dog system through vnc and open the terminal. For how to enter the system through VNC, you can watch the tutorial in Chapter 1.

7. Terminal input

   Enter the following commands in the terminal, then fill in the API keys for Tongyi Qwen and OpenRouter respectively.

x
#cm4 run
nano RaspberryPi-CM4-main/demos/API_KEY.py
​
#cm5 run
nano RaspberryPi-CM5/demos/API_KEY.py

Then enter the API-KEY of OpenRouter and the API-KEY of TONYI_KEY according to the prompts. You do not need to fill in the Spark Big Model. Then press Ctrl+S to save and Ctrl+X to exit.

Enter the following command in the terminal and fill in the OpenAI Compatible Endpoint of Tongyi Qwen.

x
#cm4 run
nano RaspberryPi-CM4-main/demos/dog_agent/dog_Track_food_api.py
​
#cm5 run
nano RaspberryPi-CM5/demos/dog_agent/dog_Track_food_api.py

Then press Ctrl+S to save and Ctrl+X to exit.

x
#cm4 run
nano RaspberryPi-CM4-main/demos/dog_agent/dog_tongyiAPI.py
​
#cm5 run
nano RaspberryPi-CM5/demos/dog_agent/dog_tongyiAPI.py

Then press Ctrl+S to save and Ctrl+X to exit.

x
#cm4 run
nano RaspberryPi-CM4-main/demos/speech_AI_food/speech_ai_food_image.py
​
#cm5 run
nano RaspberryPi-CM5/demos/speech_AI_food/speech_ai_food_image.py

Then press Ctrl+S to save and Ctrl+X to exit.

You can run the Big Model routine. The Spark Big Model key used by overseas customers has been packaged in the relevant cases and does not need to be filled in here.