Observability grid-based autonomous environment search

Patent metadata

Field	Value
Publication number	US-10754351-B2
Application number	US-201715445630-A
Country	US
Kind code	B2
Filing date	Feb 28, 2017
Priority date	Feb 28, 2017
Publication date	Aug 25, 2020
Grant date	Aug 25, 2020

Abstract

The novel technology described in this disclosure includes an example method comprising initializing an observability grid with an observation likelihood distribution for an environment being navigated by a mobile detection system, such as but not limited to a robot; searching the environment using the observability grid for an observation point; navigating, using a propulsion system, the robot to the observation point; and observing a target object from the observation point. The observability grid may include two or more spatial dimensions and an angular dimension. In some cases, the method may include sampling the environment with the robot based on the observation likelihood distribution of the observability grid.

First claim

What is claimed is:

1. A computer-implemented method for observing a target object with a robot using an observability grid, the method comprising:
initializing, using one or more processors, the observability grid with an observation likelihood distribution for an environment being navigated by the robot;
searching, using the one or more processors, the environment using the observability grid for a first observation point, the first observation point corresponding to a grid cell of the observability grid and being a physical location from which the target object is observable, the observability grid including a plurality of grid cells associated with a plurality of observation likelihood values;
navigating, using a propulsion system, the robot to the first observation point corresponding to the grid cell that has a higher observation likelihood value than other observation points corresponding to other grid cells in the observability grid;
observing, using one or more sensors of the robot, the target object from the first observation point;
updating, using the one or more processors, the observability grid based on observing the target object from the first observation point;
navigating, using the propulsion system, the robot to a second observation point;
observing, using one or more sensors of the robot, the target object from the second observation point; and
updating, using the one or more processors, the observability grid based on observing the target object from the second observation point.

2. A computer-implemented method comprising:
initializing, using one or more processors, an observability grid with an observation likelihood distribution for an environment being navigated by a mobile detection system;
searching, using the one or more processors, the environment using the observability grid for an observation point, the observation point corresponding to a grid cell of the observability grid and being a physical location from which a target object is observable, the observability grid including a plurality of grid cells associated with a plurality of observation likelihood values;
navigating, using a propulsion system, the mobile detection system to the observation point corresponding to the grid cell that has a higher observation likelihood value than other observation points corresponding to other grid cells in the observability grid; and
observing, using one or more sensors of the mobile detection system, the target object from the observation point.

3. The method of claim 2 , wherein the observability grid includes two or more spatial dimensions and an angular dimension.

4. The method of claim 2 , further comprising:
determining, using the one or more processors, to adjust a local observability based on a type of the target object and a perceptual recognition performed by the mobile detection system.

5. The method of claim 2 , wherein initializing the observability grid further comprises:
sampling, using the one or more sensors, the environment with the mobile detection system based on the observation likelihood distribution of the observability grid.

6. The method of claim 2 , wherein initializing the observability grid further comprises:
retrieving, using the one or more processors, an obstacle map;
identifying, using the obstacle map, grid cells in the observability grid that correspond to locations that are obstructed; and
reducing, using the one or more processors, observation likelihood values for corresponding grid cells in the observability grid.

7. The method of claim 6 , wherein the locations that are obstructed are unreachable by the mobile detection system.

8. The method of claim 2 , wherein the observation likelihood distribution includes one of a uniform distribution, a random initial distribution, and a predefined initial distribution.

9. The method of claim 2 , wherein initializing the observability grid includes:
processing contextual data reflecting one or more historical positions of the target object within the environment; and
increasing observation likelihood values of grid cells in the observability grid corresponding to the one or more historical positions.

10. The method of claim 2 , wherein initializing the observability grid includes:
processing contextual data reflecting one or more of a historical position, co-occurrence of animate or inanimate objects, a time, a date, a light condition, and a weather condition.

11. The method of claim 2 , further comprising:
responsive to observing the target object from the observation point, updating, using the one or more processors, the observability grid to reflect a result of the observation.

12. The method of claim 11 , wherein the observability grid is updated positively or negatively based on the result.

13. The method of claim 2 , further comprising:
determining the observation point and a path to the observation point based on an observation likelihood value associated with the observation point and one or more of a traversal speed, a distance to the observation point, and an angle of observation associated with the observation point, wherein navigating the mobile detection system to the observation point is performed using the path.

14. The method of claim 2 , wherein observing the target object from the observation point includes one of:
performing a single observation at the observation point;
performing an observation at one or more of a location before the observation point, at the observation point, and a location past the observation point; and
observing the environment for a change in one or more of a camera pose and an observed scene.

15. The method of claim 2 , wherein observing the target object from the observation point further comprises:
executing, using the one or more sensors, an observation by the mobile detection system;
determining, using the one or more processors, that the observation has detected the target object; and
updating, using the one or more processors, the observability grid based on the observation detecting the target object.

16. A system comprising:
a mobile detection system coupled to one or more sensors adapted to observe an environment, the mobile detection system including a propulsion system that moves the mobile detection system around the environment, the mobile detection system including one or more computer processors programmed to perform operations comprising:
initializing an observability grid with an observation likelihood distribution for the environment being navigated by the mobile detection system;
searching the environment using the observability grid for an observation point, the observation point corresponding to a grid cell of the observability grid and being a physical location from which a target object is observable, the observability grid including a plurality of grid cells associated with a plurality of observation likelihood values;

navigating, using the propulsion system, the mobile detection system to the observation point corresponding to the grid cell that has a higher observation likelihood value than other observation points corresponding to other grid cells in the observability grid; and
observing, using one or more sensors of the mobile detection system, the target object from the observation point.

17. The system of claim 16 , wherein the observability grid includes two or more spatial dimensions and an angular dimension.

18. The system of claim 16 , wherein the one or more computer processors are further programmed to perform operations comprising

Assignees

• Toyota Motor Co Ltd

Inventors

• Martinson Eric
• Grane Jose Capriles
• Nakano Yusuke

Classifications

• G05D1/0274 Primary

  using mapping information stored in a memory device (navigation using map-matching G01C21/30) · CPC title

• G05D1/0094

  involving pointing a payload, e.g. camera, weapon, sensor, towards a fixed or moving target · CPC title

• G05D2201/0207

  Physics · mapped topic