weixin_44492024的博客

Raspberry Pi - Robot Arm1. Robot Configuration1.1 Configure PCA96851.2 servokit libControlProject and Code1. Robot Configuration1 * Raspberry Pi processor.4 * SG90 servos.1 * PCA9685.1 * 5V(2A/4A) adaptor.3D printed arm model.1.1 Conf

Dell G3电脑配置准备准备在开始购置之前，因为不确定我的硬盘的接口型号和硬盘提供的空间大小，我先拆解了机器的背面，以做好充分的功课。内存是普通的笔记本内存硬盘有一个SATA，两个M.2的接口（其中一个为原厂装配的512G硬盘占用）...

1. Download the sourceOfficial site, but the download link is broken down, when I visited it. (We can go directly to pypi site to download the source)# tar the sourcetar xfz pycuda-VERSION.tar.gz2. Build2.1 NVCCWhen I build the project, I faced erro

Quaternion Weighted Averaging1. Objective function2. Further Derivative3. Using the rotation matrix expressionQuaternion Averaging paper1. Objective functionThis paper use the element-wise matrix difference as the objective function.Then extend the Fr

Paper read with more formula derivation: Semidefinite Programming1. Introduction1.1 SDP2. Duality2.1 Derivation2.2 Duality gap2.3 Optimal condition2.4 Example3. Primal Dual formulation4. Interior-point method4.1 Barrier function4.1.1 Derivative4.1.2 LP exa

Paper read :on the unificationof line process,outlier rejection and robust statistics1. Total variance reconstruction2. Robust Statistics3. Outlier process4. From Robust estimation to Outlier process5. GNC6. Spatial DiscontinutiesBlack M J , Rangarajan A

CVX based ICP paper read1 Banch and Bound1.1 Practical Global Optimization for Multiview Geometry1.2 Branch-and-Bound Methods for Euclidean Registration Problems2. Lagrangian Duality2.1 Convex Global 3D Registration with Lagrangian Duality2.2 Solving Quadr

Preconditioned Conjugate Gradient for Reduced Camera System in SLAM1. Introduction1.1 Linear algebra softwares1.2 Preconditioned Conjugate Gradient1.3 Prepare real SLAM data1. IntroductionTry to reproduce the article Pushing the Envelope of Modern Method

Conjugate Gradient MethodsIntroductionAlgorithmsMatlab ImplementationIntroductionCG方法一开始发明在上个世纪（1952年，Hestenes and Stiefel），作为直接求解的方法，一开始的动力是因为那时代的电脑完全没有我们现在的电脑强大。在更早的时候，甚至没有电脑存在，那种情况下使用基础的牛顿法求解一个几百...

1. 概率论大数定理2. 高斯误差的优化3. 凸优化中的凸包定理4. L2范数LJ“椭圆”是所有凸集的scale为n下的近似。L2是其他所有模在sqrt（n）下的近似。总得来说：如果在大数定理下，我们假设高斯噪音，那么最小二乘是最合适的范数选择。但是哪怕假设不成立，无论实际是什么样的误差，应该选取什么样的范数。最小二乘都可以取得一个很好的近似。也就是说，当然高斯假设下用最...

Auto-Encoding Variational Bayes from PGM概率图模型1. PGM 概率图模型1.1 Variational Inference1.2 KL divergence1.3 The Variational Lower bound1. PGM 概率图模型AVEB（Auto-Encoding Variational Bayes）的思想可以从概率图模型来理解。它是建立...

谈一下对SLAM中least square的理解1. convex optimization1.1 convex set1.2 convex function1.3 optimization problem1.4 convex optimization2. SLAM Bundle Adjustment1.5 Duality2.1 Gauss Newton's method2.2 LM法2.3 Re...

1. Lecture IntroductionProf from MIT gives us an overview of the self driving techniques.1.1 Requirements for automony vehicles.He offered us an opinion that I found very intriguing.automony vehi...

Probabilistic Graphical Modeling概率图模型学习笔记0. learning materials1. Introduction2. Representation2.1 Bayesian network2.2 Markov Random Fields2.3 Conditional Random Fields2.4 Factor Graph0. learning mate...

学习资料链接整理Convex OptimizationProbabilistic Graphical ModelingPattern Recognition and Machine LearningConvex Optimization凸优化stanford course B站链接stanford course page EE364A convex optimization Istan...

一些双目相机的总结比较（realsense，mynteye，zedmini）Realsense D435ZED MINI小觅 Mynt Eye D系列Realsense D435Realsense是我最早使用的双目深度相机，我认为realsense最大的优点就是它是市面上各种功能最齐全的深度相机，同时也是提供最完善SDK和文档的双目深度相机。总的来说，是我觉得最适合开始开发的相机。优点：...