SLAM 백엔드
06 Jul 2021 | SLAM
Camera
Local Mapping(DBow, local BA, remove outlier) -> Full Mapping(full BA, loop closure)
Lidar
Local Mapping(Scan text) - > full Mapping(Graph pose estimation, loop closure)
Summary
at backend, with database of mappoints information of keyframe, to build map
슬램 백엔드는 결국
H * delta x = b 를 푸는 문제 이다(행렬 분해를 통해 x값을 구한다.)
H is Hessian Matrix(jacobian x jacobian), Hessian is combiantion of robot pose and landmarks relatation ships.
Camera
Local Mapping(DBow, local BA, remove outlier) -> Full Mapping(full BA, loop closure)
Lidar
Local Mapping(Scan text) - > full Mapping(Graph pose estimation, loop closure)
Summary
at backend, with database of mappoints information of keyframe, to build map
슬램 백엔드는 결국
H * delta x = b 를 푸는 문제 이다(행렬 분해를 통해 x값을 구한다.)
H is Hessian Matrix(jacobian x jacobian), Hessian is combiantion of robot pose and landmarks relatation ships.
Comments