& Blogs &

This is the repository of Robotic Arm Motion Planning Simulation Based on Reinforcement Learning.

In this code, I use a self-built pybullet robot arm reinforcement learning environment to test some reinforcement learning algorithms, including DDPG, TD3, DADDPG, DATD3, and DARC, and try to let the robot arm finish three tasks: reach, push, and pick.

I use main.py to run results, the algorithms' parameters are in config.py, and use visdom to monitor the algorithms' performance. The envs deposits three self-built robot arm environments, the algo deposits test algorithms, the models deposits robot urdf file, and the utils deposits small tools for rl-learning.

This is the blog of Plotclock, as for The Astral Movement Clock.

In this repository, I use 3D printing to realize the shell design of the luminous clock, and use arduino to program it, so that the servo can erase the luminous numbers, which is very fun and cute.

Different from the original three servo design, the UV light is used instead of the traditional oil-based pen, which reduces the fragrance of pen and ink, but adds a lot of charming light pollution.

This is the blog of Smart Car Upper End for The 17th National College Student Smart Car Competition.

In this repository, I used pyqt5 to build the entire host computer framework, which can realize three functions of offline debugging, online debugging, and real-time debugging, helping us to quickly debug pictures and speed up image processing.

This is the blog of WeChat Applet For Triboelectric Nanogenerator(TENG)

This WeChat applet needs to be used with the smart tires developed by us. It can analyze the electrical signals generated by the smart tires when rolling, and transmit information such as vehicle speed, tire pressure, and skid resistance to the applet for display, which can remind driving, which can remind people pay attention to road safety.

一套基于TCP协议的主从模式聊天系统，包括单对单、单对多的文字信息通信.

项目在Server端实现了多Client的自动分配Select机制，从而连接多个client分别通信，并通过SQLite数据库实现用户信息存取以及日志记录。在client端采用QT制作GUI界面，优化用户体验，构成了一套较完整的架构体系。

In this project, we design an intelligent domestic waste sorting device based on raspberry PI. The mechanical structure of the device mainly includes self-made trash can, raspberry PI, two degree of freedom steering gear, sorting hopper, etc. put the garbage to be sorted into the sorting hopper, read the video stream through Open CV for identification. After successful identification, under the control of raspberry pie, the two degree of freedom steering gear can accurately drop the garbage, so as to realize the independent dropping of domestic garbage. Among them, the recognition algorithm is realized by neural network model and infrared tube. After the infrared tube is judged, the camera is sent to the model for comparison.