Thanks for visiting my blog

Here I exhibit some of my projects and research, and books I love. I’m interested in Deep-Learning, Robotics and Psychology. The primary objective of this blog is to encapsulate the myriad of technical challenges and insights I encounter on my educational journey, thereby aiding others. Furthermore, it serves as a practical application of the Feynman Technique, enhancing my own capacity for knowledge dissemination.

Should there be any inaccuracies or aspects of the blog that elicit confusion, please do not hesitate to contact me via email at your earliest convenience.

Diffusion Model 4 Protein

​ Inverse protein folding is challenging due to its inherent one-to-many mapping

characteristic, where numerous possible amino acid sequences can fold into a single,

identical protein backbone. This task involves not only identifying viable sequences

but also representing the sheer diversity of potential solutions. This article is the overview of the technology utilized by Graph Denoising Diffusion for Inverse Protein Folding.

ResNet

​ 残差学习就是改变传统的网络结构，将层间传输的数据转换为F(x)+x的形式。优势在于梯度不会随着导数的相乘而快速趋近于0，使得网络在达到Plateau后继续下降收敛。

单片机开发 | MCU Development

A microcontroller (MCU for microcontroller unit) is a compact integrated circuit designed to govern a specific operation in an embedded system. Typically, microcontrollers include an application-specific processor, memory, and input/output peripherals on a single chip.

By offering a customizable mix of read/write memory, program memory, and processing capability, microcontrollers make it economical to digitally control even more devices and processes.

Figure 1 depicts the circuit diagram we constructed within the simulation software to verify the rationality of the design.

Embedded System

Smart vehicle

Building on the hardware development knowledge acquired earlier, we completed an intelligent car based on computer vision (CV) and a two-wheeled balancing vehicle using a gyroscope. These systems are significantly more complex, deepening our understanding of control principles such as speed feedback, dual-loop control, and fuzzy PID control, as well as computer vision and ARM embedded development, including GPIO and nested interrupts.

多智能体 | Multi-Agents

寻路算法

DWA is a local planner used to control a mobile robot by selecting the optimal velocity and steering angle based on the robot’s current state and its goal. It takes into account the robot’s dynamics and the environment to avoid collisions while moving towards the target.

机器人课程总结 | Robotics

六关节机械臂ROS仿真与实验

In industrial applications, robotic arms with multiple joints possess a high degree of freedom, necessitating the spatial modeling of nodes and linkages, known as the DH (Denavit-Hartenberg) parameter table. Euler coordinate transformations are applied to convert joint space into Cartesian coordinate space to accomplish specific operational tasks. Figure 1 showcases a classic 6DOF (Degrees of Freedom) robot and a recognition and grasping task based on GAZEBO machine vision.

记录一次痛苦安装 CUDA 及 cuDNN，希望以后不幸需要再装时能轻松些。

人工智能课上学了点遗传算法，记录搓的一段比较简单的实现代码，实现求解
$$
f(x)=xe^{sin(2\pi x)}
$$
在区间 $[0,5]$ 的最大值
(参数和算法设计不一定合理)

转载，原文地址:https://mysteriouspreserve.com/blog/2022/11/06/WSL2-MySQL/

在 WSL2 上安装的 MySQL 无法在 Windows 上直接由 127.0.0.1 访问，在此记录配置 MySQL 远程连接使得 Windows 的 SQL 管理软件可以直接管理 WSL2 里的 MySQL的方法。

数字控制器的离散化设计步骤

计算机控制系统框图如下图所示：

可得其闭环脉冲函数为：
$$
\Phi(z)=\frac{D(z) G(z)}{1+D(z) G(z)}
$$
其中，$G(z)=Z[H(s)G_c(s)]$，为控制对象加上零阶保持器后的离散化。

反推可知：
$$
D(z)=\frac{\Phi(z)}{G(z)(1-\Phi(z))}=\frac{\Phi(z)}{G(z) \Phi_e(z)}
$$
设计最少拍控制器，通常根据性能要求和约束条件确定所需的$\Phi(z)$和$\Phi_e(z)$，再得到$D(z)$。