Localizing and mapping platform
View Patent ↗Implementations generally relate to localizing and mapping. In one implementation, a method includes determining one or more map points in a point cloud space, where the point cloud space corresponds to a real physical environment. The method further includes determining movement information of one or more of the map points, where the determining of the movement information is performed by a neural network. The method further includes determining a self-position in the point cloud space based on the determined movement information.
1. A method comprising:
determining one or more map points in a point cloud space, wherein the point cloud space corresponds to a real physical environment;
determining movement information of one or more of the map points, wherein the determining of the movement information is performed by a neural network, and wherein, to determine the movement information, the method further comprises:
determining first positional information associated with a first image frame;
determining second positional information associated with a second image frame; and
determining one or more differences between the first positional information and the second positional information; and
determining a self-position in the point cloud space based on the determined movement information.
2. The method of claim 1 , wherein the movement information comprises two-dimensional changes in position in a two-dimensional window.
3. The method of claim 1 , wherein the movement information comprises two-dimensional changes in rotation in a two-dimensional window.
4. The method of claim 1 , wherein the movement information comprises three-dimensional changes in the point cloud space.
5. The method of claim 1 , wherein the determining of the self-position is performed by the neural network.
6. The method of claim 1 , wherein input to the neural network comprises two-dimensional digital images that are digitized image frames from a camera.
7. The method of claim 1 , wherein input to the neural network comprises inertial information.
8. The method of claim 1 , wherein the neural network is a recurrent neural network.
9. The method of claim 1 , wherein the determining of movement information is based on training of the neural network.
10. The method of claim 1 , wherein the determining of movement information is based on training of the neural network, and wherein the training of the neural network includes providing ground truth to the neural network by a hardware device.
11. A non-transitory computer-readable storage medium carrying program instructions thereon, the instructions when executed by one or more processors are operable to cause the one or more processors to perform operations comprising:
determining one or more map points in a point cloud space, wherein the point cloud space corresponds to a real physical environment;
determining movement information of one or more of the map points, wherein the determining of the movement information is performed by a neural network, and wherein, to determine the movement information, the instructions when executed are further operable to cause the one or more processors to perform operations comprising:
determining first positional information associated with a first image frame;
determining second positional information associated with a second image frame; and
determining one or more differences between the first positional information and the second positional information; and
determining a self-position in the point cloud space based on the determined movement information.
12. The computer-readable storage medium of claim 11 , wherein the movement information comprises two-dimensional changes in position in a two-dimensional window.
13. The computer-readable storage medium of claim 11 , wherein the movement information comprises two-dimensional changes in rotation in a two-dimensional window.
14. The computer-readable storage medium of claim 11 , wherein the movement information comprises three-dimensional changes in the point cloud space.
15. The computer-readable storage medium of claim 11 , wherein the determining of the self-position is performed by the neural network.
16. The computer-readable storage medium of claim 11 , wherein input to the neural network comprises two-dimensional digital images that are digitized video frames from a camera.
17. A system comprising:
one or more processors;
logic encoded in one or more non-transitory computer-readable storage media for execution by the one or more processors and when executed operable to cause the one or more processors to perform operations comprising:
determining one or more map points in a point cloud space, wherein the point cloud space corresponds to a real physical environment;
determining movement information of one or more of the map points, wherein the determining of the movement information is performed by a neural network, and wherein, to determine the movement information, the logic when executed are further operable to cause the one or more processors to perform operations comprising:
determining first positional information associated with a first image frame;
determining second positional information associated with a second image frame; and
determining one or more differences between the first positional information and the second positional information; and
determining a self-position in the point cloud space based on the determined movement information.
18. The system of claim 17 , wherein the movement information comprises two-dimensional changes in position in a two-dimensional window.