HeYijia

Results 25 comments of HeYijia

Yes, you are right. In addition, you can use the [simulation code](https://github.com/MegviiRobot/CamLaserCalibraTool/blob/master/main/calibr_simulation.cpp#L81) I provided to verify your conclusion. You will find a nullspace base that shows the system has an...

1. 如果你是用的 kalibr 的 pdf, tag_type 就选择 1, 注释已经说明了。 2. 0.165 是我自己打印的 apriltag 的尺寸,不是 kalibr tag 的尺寸,请自行修改。

1. 不应该每次执行结果都不一致哈。2. 不需要全部,只需要几个就行,apritag 最好别离相机太远,1-2m。3. 坐标系是 SLAM 中通用的坐标系。相机 z 轴是从相机光心指向图像平面,x 轴是图像平面中心水平从左向右的向量,y 轴是图像平面中心竖直从上往下的向量。 激光水平放在地面,正前方是坐标系的 x 轴,垂直地面向上是 z 轴,y 轴是水平从右往左。

I also faced the similar problem when I using the dvo package. Did anyone figure out? Many thanks in advance.

误差计算没有问题。你可以把 imu 频率降低一点如 100,50,10 hz,让问题突出一点再对比下精度下。看看是不是你实现的不对。

Since the line detector is time-consuming, our plvio cannot run in real-time. You can collect data and run it with offline mode.

需要修改参数,可以先添加特征点数目数目。

先定位下问题是检测不到特征点还是特征点容易跟丢? 图像太大,像素分辨率高的情况下,你可以修改下金字塔跟踪的层数,以及修改下特征跟踪丢失的阈值。

I am sorry for that the semi-dense-orbslam code is ugly. At that time, I just want to learn the paper of semi-dense-orbslam so that I only spend two weeks to...