shuttworth / Record_Issues_For_LVI-SAM_detailed_comments

7 stars 1 forks source link

用百度网盘中的 dataset: Service_Center1.bag 去跑代码,但效果并不理想 #3

Open LiXiaoQiang123z opened 2 years ago

LiXiaoQiang123z commented 2 years ago

我用的是 electech6/LVI-SAM_detailed_comments 中的代码, branch = master 。 去跑Service_Center1.bag,但效果并不理想。 是因为我拉取的branch不对吗? 那需要拉取哪个分支? 非常感谢

shuttworth commented 2 years ago

First, you said the result was not satisfactory, do you mean the map built is unsatified? Second, because the imu timestamp is not accurate, the effect is not very good. At present, a new batch of equipment has arrived, and will re-record data packets after assembling Last, you can try M2DGR datasets , branch = M2DGR_branch.

LiXiaoQiang123z commented 2 years ago

非常感谢回复解答。 针对第一个: First,建图过程中的估计漂移抖动剧烈。 且一直提示 [ERROR] [1665921575.994378461]: VINS big IMU acc bias estimation XXX, restart estimator! [ERROR] [1665921575.994407782]: VINS failure detection! [ERROR] [1665921575.995304465]: VINS system reboot! rviz中建图结果显示如下: 图片

针对第二个:确实,vins中一直提示IMU的acc值有问题, 可能就是IMU时间戳对齐等的影响吧。 针对第三点:我目前正在下载该M2DGR数据集,非常感谢提示。

需要在占用下您的时间,有个疑问:params_daheng.yaml 文件中的 lidar_to_camry=-0.8 ,是为什么呢? 如果说lidar_to_cam对应的坐标变换是 lidar to imu的话, 平移关系t是一致的,旋转关系ry不应该也是为0吗? 为什么会是 -0.8呢? 还是说lidar_to_cam不完全是lidar to imu ,而是lidar to world , 而这个“world”与imu的坐标系 位置一致,但坐标轴有所不同呢? 还是说是别的原因呢? 非常感谢。