320M5 Deepseek case

Function introduction: This case uses Baidu online speech recognition API to convert user speech into text, and passes the text command to the locally deployed DeepSeek-R1 large language model. DeepSeek-R1 arranges predefined function functions according to the command content and returns a structured JSON dictionary. The subsequent script parses the JSON data, combines visual technologies such as OpenCV, YOLO target detection and OCR text recognition, and guides the robotic arm to complete the grasping operation

1 Case operation configuration requirements

Requirements Configuration
Computer type Laptop
CPU 12th Gen Intel(R) Core(TM) i7-12700H 2.30 GHz
GPU NVIDIA GeForce RTX 3060 or above
Memory 16.0 GB
Hard disk 128G available space
System Windows 10 and above
Network requirements Able to access the Internet normally
Robotic arm 320M5
Python Version 3.9 or above

2 Hardware installation

Manually turn the J6 joint to the zero scale line Image

Be sure to install the camera and force control gripper flange to the end of the robot arm according to the figure below. During installation, be careful not to move the J6 zero scale line, otherwise the robot arm will not be able to grasp normally Image

Install the force control gripper on the flange Image

Connect the M8 aviation plug cable to the end interface of the robot arm and the force control gripper interface respectively Image

Finally, connect the camera cable to the computer USB port

3 ollama installation

Download official website: https://ollama.com/

4 deepseek Model installation

Open the terminal, enter the following command and press Enter to install

ollama run deepseek-r1:7b

After installation, restart the computer

5 Baidu API creation

Register Baidu AI Open Platform account First, visit Baidu AI Open Platform, register and log in to your account.

Image

Create an application and obtain APP_ID, API Key and Secret Key In the Baidu AI Open Platform console, create a new application and select the "Speech Recognition" service. After completing the application creation, you will get a pair of API Key and Secret Key as well as APP_ID for subsequent API call authentication.

Image

Image

Image

Image

Image

Image

5 Source code acquisition

Download address: https://github.com/elephantrobotics/320_deepseek_demo

Download the compressed package and solve it before deploying the environment

Image

Dependency library installation

pip install -r requirements.txt

Image

API_Key fill in

In the API_Key.py script file, fill in your Baidu APP_ID, API Key, Secret Key

7 Case reproduction

Run deepseek_demo.py, press the Enter key, and the recording will start

Notes When recording, it is necessary to do it in a quiet environment. When converting voice to text, the conversion may be incorrect. You can wait until the robot arm completes the action, and then press the Enter key again to speak the command

7.1 Mobile display

Voice command words:

向前移动50毫米,向后移动40毫米,向左移动30毫米,向右移动60毫米。

7.2 Color recognition

The colored wooden blocks should be placed neatly, and a certain distance should be kept between the wooden blocks, otherwise the robot arm cannot complete the grasping normally , please refer to the following figure for placement

Picture

Voice command words:

先把番茄一样颜色的木块拿走,再把和香蕉一样颜色的木块拿走

7.3 Fruit recognition

The fruit models should be placed neatly and a certain distance should be kept between them, otherwise the robot arm cannot grasp them normally. Please refer to the following figure for placement. If you want to grasp a banana, place it vertically

Image

Voice command:

我现在有点口渴,帮我拿个橙子。
我现在有点饿,帮我拿个苹果。

Note: Currently, fruit recognition only supports recognition of apples, oranges, and bananas

7.4 Drug recognition

The medicine boxes to be grasped must be placed neatly and vertically, without too much angle deviation, and a certain distance should be kept between the medicine boxes, otherwise the robot arm cannot grasp them normally. Please refer to the following figure for placement.

图片

Notes: Currently, the medicine box recognition only supports the recognition of 999 Ganmao Ling Granules and Yunnan Baiyao Band-Aids. If you want to reproduce, please prepare this medicine box in advance

Voice command words:

我有点流鼻涕和鼻塞,帮我取个治疗这个症状的药。

7.5 Case video

Video link: https://www.bilibili.com/video/BV1coMczkEGd/?spm_id_from=333.337.search-card.all.click&vd_source=672e3f7240eaaca210b45e7c033dc45f

results matching ""

    No results matching ""