If i use an optical flow sensor on an rover (similar to what is shown in this video with ADNS3080 mouse sensor) for odometry,
i can obtain not only linear movement but also displacement (along two axes on a plane) as well as vertical speed from a variometer (barometer), thus obtaining a full-fledged three-dimensional velocity vector. If add them in vector wheel velocity to the calculation:
can it improve the lidar location estimate? Is it possible, based on this three-dimensional velocity vector, if it turns out to be accurate enough, to carry out dead reckoning at moments when the lidar does not see landmarks around it, in a planar scene?
I noticed that the wheel encoder only uses linear speed on one axis:
https://github.com/ZikangYuan/liw_oam/blob/85fc480c66453228e064c112dd405c07b975d200/src/lioOptimization.cpp#L1323
If i use an optical flow sensor on an rover (similar to what is shown in this video with ADNS3080 mouse sensor) for odometry,
i can obtain not only linear movement but also displacement (along two axes on a plane) as well as vertical speed from a variometer (barometer), thus obtaining a full-fledged three-dimensional velocity vector. If add them in vector wheel velocity to the calculation:
https://github.com/ZikangYuan/liw_oam/blob/85fc480c66453228e064c112dd405c07b975d200/src/lioOptimization.cpp#L1324
can it improve the lidar location estimate? Is it possible, based on this three-dimensional velocity vector, if it turns out to be accurate enough, to carry out dead reckoning at moments when the lidar does not see landmarks around it, in a planar scene?