IMU数据滤波
[TOC]
一阶低通滤波
一阶低通滤波,又叫 一阶惯性滤波,其算法公式为:
\[Y_k = \alpha X_k + (1-\alpha) Y_{k-1}\]
根据安卓开发者文档中提供的低通滤波算法:
// https://developer.android.com/guide/topics/sensors/sensors_motion#java
public void onSensorChanged(SensorEvent event){
// In this example, alpha is calculated as t / (t + dT),
// where t is the low-pass filter's time-constant and
...
Ego-motion vs Odometry vs Pose Estimation
[TOC]
Ego-motion
Ego-motion is defined as the 3D motion of a system (ex camera) within an environment.
Visual Odometry
Odometry is generally used to refer to an elapsed or relative position and orientation and, because it is not absolute, is prone to drift over time.
Visual odometry in general is the estimation of ego-m...
Bundle Adjustment (BA) in vSLAM or SFM
[TOC]
Overview
Bundle adjustment (wikipedia)
BA is a key ingredient of Structure and Motion Estimation (SaM), almost always used as its last step
It is an optimization problem over the 3D structure and viewing parameters (camera pose, intrinsic calibration, radial distortion parameters), which are simultaneous...
EKF v.s. ESKF or Direct v.s. Indirect KF
[TOC]
Overview
EKF and ESKF:
EKF: Extended Kalman Filter –> Direct Kalman Filter
ESKF: Error State Kalman Filter –> Indirect Kalman Filter
Estimator State Types:
True State
Nominal State
Error State
ESKF Advantages:
The orientation error-state is minimal (i.e., it has the same number of parameters as degrees of...
Data Association in MSCKF
[TOC]
MSCKF
typedef
std::map<FeatureIDType, Feature, std::less<int>,
Eigen::aligned_allocator<std::pair<const FeatureIDType, Feature> > > MapServer;
struct Feature {
FeatureIDType id;
Eigen::Vector3d position;
std::map<StateIDType, Eigen::Vector4d, std::less<StateIDType>,
Eigen::aligned_allocator...
VINS-Mono LoopClosure
[TOC]
Overview
忽略掉了 shift to base frame:w_t_vio, w_r_vio
loop closure 初始化
PoseGraph 设置
设置参数
启动线程 PoseGraph::optimize4DoF 或 PoseGraph::optimize6DoF
加载 词典文件
启动线程 SystemROS::loop_closing
线程 loop_closing
获取同步的图像、位姿、2D与3D点信息
img
pose
point_id, point_3d, point_2d_uv, ...
the Inflated IMU Noise Params in SLAM
ref: https://vision.in.tum.de/data/datasets/visual-inertial-dataset
图像分析之ORB特征
[TOC]
Overview
Oriented FAST + Rotated BRIEF
Feature/Corner Detector
FAST方法 – 定位特征点坐标
矩方法 – 特征点方向
非极大值抑制 – 特征点响应值(分数)
图像金字塔 – 特征点具有 尺度不变性
Descriptor
BRIEF – 特征点描述子
特征点方向 – 描述子具有 旋转不变性
oFAST: FAST Keypoint Orientation
Multiscale Image Pyramid
level: 8
scale: 1.2
down sample: bilinear ...
100 post articles, 13 pages.