Robot self-position identification system and self-position identification method

• US 7,263,412 B2
• Filed: 11/27/2002
• Issued: 08/28/2007
• Est. Priority Date: 11/30/2001
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Fukuchi, Masaki, Gutmann, Steffen
• Primary Examiner(s): Black; Thomas
• Assistant Examiner(s): Marc; McDieunel
• Application Number:
  US10/470,456
• Publication Number:
  US 20040249504A1
• Time in Patent Office: 1,735 Days
• Field of Search: 700/245
• US Class Current: 700/245
• CPC Class Codes: G05D 1/0231 using optical position detecting means position-fixing by using electromagnetic waves other than radio waves, e.g. optical position detecting means G01S5/16
• CPC Class Codes: G05D 1/0246 using a video camera in combination with image processing means
• CPC Class Codes: G05D 1/0272 comprising means for registering the travel distance, e.g. revolutions of wheels measuring distance traversed on the ground by vehicles, e.g. using odometers G01C22/00
• CPC Class Codes: G05D 1/0274 using mapping information stored in a memory device navigation using map-matching G01C21/30
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59

Abstract

A self-localization system uses both of a global search apparatus, which is based on grid-based Markov localization, and a local search apparatus, which uses an extended Kalman filter. If an observation result by the global search apparatus is valid, then updating of an observation result by the local search apparatus is permitted, but if the observation result is not valid, then updating of the observation result by the local search apparatus is not performed. On the other hand, if the observation result by the local search apparatus is valid, then a state of the local search is outputted, but if the observation result is not valid, then the local search apparatus is re-initialized. Accordingly, the self-localization of a robot can be performed based on sensor information of the robot and motion information performed by the robot in an environment including artificial landmarks.

Claims

1. A self-localization system for a robot that includes a moving mechanism and can travel in an environment comprising:

external observation means for observing a landmark provided in the environment;

internal observation means for observing motion information of said robot;

first search means for outputting a localization solution at a comparatively high speed based on a result of the external observation and a result of the internal observation;

second search means for outputting a localization solution with a comparatively fine accuracy based on the external observation result and the internal observation result; and

control means for controlling the operation of said second search means in response to the output result of said first search means.

2. A self-localization system for a robot according to claim 1, wherein

said first search means applies Markov localization to store the position of said robot in the environment as a self-localization probability density distribution on discrete grids, and estimates the self-localization based on a relative position of said robot from the landmark in response to the observation result of the landmark by said external observation means to update the self-localization probability density distribution, updates the self-localization probability density based on the observation result of the motion information of said robot by said internal observation means, and outputs that one of said grids having the highest value of the self-localization probability density distribution at each time as an estimation result of the self-localization.

3. A self-localization system for a robot according to claim 2, wherein

said second search means applies an extended Kalman filter including a state model, which defines a relationship between the motion information of said robot by said internal observation means and the self-localization, and an observation model, which defines a relationship between the self-localization and the observation information of the landmark by said external observation means, to estimate the self-localization based on the motion information of said robot by said internal observation means and the observation information of the landmark by said external observation means.

4. A self-localization system for a robot according to claim 3, wherein

said control means compares the estimation result of the self-localization by said second search means with the estimation result of the self-localization by said first search means to verify the validity of the estimation result of the self-localization by said second search means and outputs the search result of the self-localization by said second search means as the self-localization result in response to a fact that the evaluation satisfies a predetermined reference, but re-initializes said second search means in response to another fact that the evaluation does not satisfy the predetermined reference.

5. A self-localization system for a robot according to claim 4, wherein

said control means evaluates that the output result of said second search means is not valid when the similarity of the self-localization estimation results of said first search means and said second search means is low.

6. A self-localization system for a robot according to claim 2, wherein

said control means evaluates the validity of the external observation result by said external observation means based on a probability with which the landmark is observed by said external observation means at the self-localization estimated by said first search means and limits inputting the external observation result to said second search means in response to a fact that the evaluation does not satisfy a predetermined reference.

7. A self-localization system for a robot according to claim 1, wherein

said control means evaluates the validity of the external observation result by said external observation means based on the search result by said first search means and limits inputting the external observation result to said second search means in response to a fact that the evaluation does not satisfy a predetermined reference.

8. A self-localization system for a robot according to claim 1;

wherein

said control means verifies the validity of the search result by said second search means and outputs the search result as the self-localization result in response to a fact that the evaluation satisfies a predetermined reference, but re-initializes said second search means in response to another fact that the evaluation does not satisfy the predetermined reference.

9. A self-localization system for a robot according to claim 1, wherein

said external observation means includes image inputting means for capturing the landmark and image recognition means for recognizing the landmark and the distance to the landmark based on an input image from said image inputting means.

10. A self-localization system for a robot according to claim 1, wherein

said external observation means includes a range finder for measuring the distance to the landmark.

11. A self-localization system for a robot according to claim 1, wherein

the landmark has visually discernible identification information formed on a surface thereof.

12. A self-localization method for a robot that includes a moving mechanism and can travel in an environment comprising:

an external observation step of observing a landmark provided in the environment;

an internal observation step of observing motion information of said robot;

a first search step of outputting a localization solution at a comparatively high speed based on a result of the external observation and a result of the internal observation;

a second search step of outputting a localization solution with a comparatively fine accuracy based on the external observation result and the internal observation result; and

a control step of controlling the operation of the second search means in response to the output result of the first search means.

13. A self-localization method for a robot according to claim 12, wherein

the first search step applies Markov localization to store the position of said robot in the environment as a self-localization probability density distribution on discrete grids, and estimates the self-localization based on a relative position of said robot from the landmark in response to the observation result of the landmark by the external observation step to update the self-localization probability density distribution, updates the self-localization probability density based on the observation result of the motion information of said robot by the internal observation step, and outputs that one of said grids having the highest value of the self-localization probability density distribution at each time as an estimation result of the self-localization.

14. A self-localization method for a robot according to claim 13, wherein

the control step evaluates the validity of the external observation result by the external observation step based on a probability with which the landmark is observed by the external observation step at the self-localization estimated by the first search step and limits inputting the external observation result to the second search step in response to a fact that the evaluation does not satisfy a predetermined reference.

15. A self-localization method for a robot according to claim 12, wherein

the second search step applies an extended Kalman filter including a state model, which defines a relationship between the motion information of said robot by the internal observation step and the self-localization, and an observation model which, defines a relationship between the self-localization and the observation information of the landmark by the external observation step, to estimate the self-localization based on the motion information of said robot by the internal observation step and the observation information of the landmark by the external observation step.

16. A self-localization method for a robot according to claim 15, wherein

the control step compares the estimation result of the self-localization by the second search step with the estimation result of the self-localization by the first search step to verify the validity of the estimation result of the self-localization by the second search step and outputs the search result of the self-localization by the second search step as the self-localization result in response to a fact that the evaluation satisfies a predetermined reference, but re-initializes the second search step in response to another fact that the evaluation does not satisfy the predetermined reference.

17. A self-localization method for a robot according to claim 16, wherein

the control means evaluates that the output result of the second search step is not valid when the similarity of the self-localization estimation results of the first search step and the second search step is low.

18. A self-localization method for a robot according to claim 12, wherein

the control step evaluates the validity of the external observation result by the external observation step based on the search result by the first search step and limits inputting the external observation result to the second search step in response to a fact that the evaluation does not satisfy a predetermined reference.

19. A self-localization method for a robot according to claim 12, wherein

the control step verifies the validity of the search result by the second search step and outputs the search result as the self-localization result in response to a fact that the evaluation satisfies a predetermined reference, but re-initializes the second search step in response to another fact that the evaluation does not satisfy the predetermined reference.