Method of building map of mobile platform in dynamic environment
Disclosed herein is a method of building a map of a mobile platform moving in a dynamic environment and detecting an object using a 3D camera sensor, e.g., an IR TOF camera sensor, for localization. A localization technology to separate and map a dynamic object and a static object is applied to a mobile platform, such as an unmanned vehicle or a mobile robot. Consequently, the present method is capable of accurately building map information based on the static object in a dynamic environment having a large number of dynamic objects and achieving a dynamic object avoidance or chasing function using position information acquired to build the map.
1. A method of building a map of a mobile platform, comprising:
acquiring, by a processor, a three-dimensional (3D) image of an environment in which the mobile platform moves;
extracting feature points from the 3D image to estimate motion states of the respective feature points;
clustering the feature points based on the motion states of the respective feature points to divide the feature points into a dynamic map and a static map; and
building the mobile platform map based on the static map.
2. The method according to claim 1 , wherein the 3D image is a disparity map including 3D information of a dynamic object and a static object in a mixed state.
3. The method according to claim 2 , wherein the disparity map is acquired by a 3D camera sensor.
4. The method according to claim 1 , wherein the extraction of the feature points is achieved using a volume descriptor of the 3D image.
5. The method according to claim 1 , wherein the estimating the motion states of the feature points comprises estimating positions and angular velocities of the feature points using a Joint Probabilistic Data Association Filter (JPDAF).
6. The method according to claim 5 , wherein the clustering the feature points comprises determining whether the feature points belong to a dynamic object or a static object using the positions and angular velocities of the feature points and clustering the feature points pre-estimated as the same object into a dynamic cluster or a static cluster.
7. The method according to claim 6 , wherein only the feature points included in the static cluster are separated and registered in the static map.
8. The method according to claim 7 , further comprising calculating a position of the mobile platform using only the feature points registered in the static map and building the map of the mobile platform with the dynamic object removed according to the position of the mobile platform.
9. The method according to claim 6 , wherein only the feature points included in the dynamic cluster are separated and registered in the dynamic map.
10. The method according to claim 9 , further comprising creating a dynamic object avoidance path based on the dynamic map.
11. The method according to claim 10 , wherein the creating the dynamic object avoidance path comprises extracting an average velocity of the dynamic cluster registered in the dynamic map to pre-estimate a path of the dynamic cluster and creating the dynamic object avoidance path based on the path of the dynamic cluster.
12. A method of building a map for a movable unit in a dynamic environment, comprising:
acquiring, by a processor, a three-dimensional (3D) image of an environment having a dynamic object;
extracting feature points from the 3D image to estimate motion states of the respective feature points;
clustering the feature points based on the motion states of the respective feature points to divide the feature points into a dynamic map and a static map; and
building the movable unit map based on the static map.
13. The method according to claim 12 , wherein the extraction of the feature points is achieved using a volume descriptor of the 3D image.
14. The method according to claim 12 , wherein the estimating the motion states of the feature points comprises estimating positions and angular velocities of the feature points using a Joint Probabilistic Data Association Filter (JPDAF).
15. The method according to claim 14 , wherein the clustering the feature points comprises determining whether the feature points belong to the dynamic object or a static object using the positions and angular velocities of the feature points and clustering the feature points pre-estimated as the same object into a dynamic cluster or a static cluster.
16. The method according to claim 15 , wherein only the feature points included in the static cluster are separated and registered in the static map.
17. The method according to claim 16 , further comprising building the movable unit map with the dynamic object removed using only the feature points registered in the static map.
18. The method according to claim 15 , wherein only the feature points included in the dynamic cluster are separated and registered in the dynamic map.
19. The method according to claim 18 , further comprising creating a dynamic object avoidance path based on the dynamic map.
20. The method according to claim 19 , wherein the creating the dynamic object avoidance path comprises extracting an average velocity of the dynamic cluster registered in the dynamic map to pre-estimate a path of the dynamic cluster and creating the dynamic object avoidance path based on the path of the dynamic cluster.
21. A method of creating a dynamic object avoidance path of a mobile platform, comprising:
acquiring, by a processor, a three-dimensional (3D) image of an environment in which the mobile platform moves;
extracting feature points from the 3D image to estimate motion states of the respective feature points;
clustering the feature points based on the motion states of the respective feature points to divide the feature points into a dynamic map and a static map; and
creating the dynamic object avoidance path based only on the dynamic map.
22. The method according to claim 21 , wherein the clustering the feature points comprises determining whether the feature points belong to a dynamic object or a static object using positions and angular velocities of the feature points and clustering the feature points pre-estimated as the same object into a dynamic cluster or a static cluster.
23. The method according to claim 22 , wherein only the feature points included in the dynamic cluster are separated and registered in the dynamic map, and wherein the creating the dynamic object avoidance path comprises extracting an average velocity of the dynamic cluster registered in the dynamic map to pre-estimate a path of the dynamic cluster and creating the dynamic object avoidance path based on the path of the dynamic cluster.