OBJECT TRACKING USING LINEAR FEATURES

• US 20100034422A1
• Filed: 08/06/2008
• Published: 02/11/2010
• Est. Priority Date: 08/06/2008
• Status: Active Grant

• Current Assignee: Toyota Jidosha Kabushiki Kaisha (Toyota Motor Corporation)
• Original Assignee: Toyota Motor Engineering & Manufacturing North America Incorporated (Toyota Motor Corporation)
• Inventors: Dolgov, Dmitri A., James, Michael Robert
• Granted Patent: US 8,705,792 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments

Abstract

A method of tracking objects within an environment comprises acquiring sensor data related to the environment, identifying linear features within the sensor data, and determining a set of tracked linear features using the linear features identified within the sensor data and a previous set of tracked linear features, the set of tracked linear features being used to track objects within the environment.

Claims

1. A method of tracking objects within an environment, the method comprising.

acquiring sensor data related to the environment;

identifying present linear features within the sensor data;

tracking previous linear features from a previous set of tracked linear features to the present linear features; and

creating a set of tracked linear features, the set of tracked linear features being used to track objects within the environment,

the set of tracked linear features comprising;

present linear features tracked from previous linear features;

present linear features not tracked from previous linear features; and

previous linear features not tracked to present linear features.

2. The method of claim 1, further comprising:

determining mapping probabilities for mapping previous linear features to each present linear feature;

identifying a highest mapping probability; and

tracking the previous linear feature to the present linear feature if the highest mapping probability is greater than a threshold value.

3. The method of claim 2, the mapping probabilities being expressed in terms of at least one parameter selected from spatial displacement, stretching, and reorientation.

4. The method of claim 1, each linear feature within the present set of tracked linear features having an feature probability,

the feature probability including a contribution correlated with linearity.

5. The method of claim 4, the feature probabilities of previous linear features within the set of tracked linear features being less than probability values used in the previous set of tracked linear features.

6. The method of claim 1, the sensor data including range data, the line features being line segments or planes.

7. The method of claim 1, the method being used to assist automated navigation of a vehicle through the environment.

8. A method of tracking objects within an environment, the method comprising:

identifying present linear features within the environment using sensor data obtained from the environment;

identifying mappings of previous tracked linear features to the present linear features;

determining a feature probability for each present linear feature; and

creating a set of present tracked linear features including the present linear features,

the present tracked linear features further including unmapped previous linear features not mapped to present linear features, the feature probability of unmapped previous linear features being reduced,

the set of tracked linear features and mappings being used to track objects within the environment.

9. The method of claim 8, an unmapped previous linear feature being removed from the set of present tracked linear features if the associated feature probability is less than a threshold value.

10. The method of claim 8, wherein mapping previous tracked linear features to the present linear features comprises determining mapping probabilities for a plurality of possible mappings, and selecting a mapping with a mapping probability greater than a threshold level.

11. The method of claim 10, the mapping probability being expressed in terms of at least one parameter selected from spatial displacement, stretching, and reorientation.

12. The method of claim 8, the sensor data including range data obtained using a range sensor, the range sensor being selected from a group of sensors consisting of a lidar sensor, a radar sensor, and a sonar sensor.

13. A method of tracking objects within an environment, the method comprising:

obtaining sensor data from the environment, the sensor data including range data;

identifying linear features within the environment using the sensor data; and

tracking objects within the environment by tracking the linear features,

wherein the linear features are identified within the sensor data using a locality-based neighborhood calculation including;

determining a normal vector for each sensor data point using sampled sensor data around each sensor data point; and

clustering proximate sensor data points having similar normal vectors.

14. The method of claim 13, wherein the sampled sensor data around each sensor data is selected using an adaptive local distance function.

15. The method of claim 13, linear features formed by clustering proximate sensor data points being further extended by:

including sensor data points having positions consistent with an extension of the linear feature into the linear feature.

16. The method of claim 13, wherein the sensor data represents a two-dimensional array of range data, and the linear features are line segments.

17. The method of claim 13, wherein the sensor data represents a three-dimensional array of range data, and the linear features are planes.