Method and apparatus to localize an autonomous vehicle using convolution
View Patent ↗Aspects of the disclosure relate generally to an autonomous vehicle accessing portions of a map to localize itself within the map. More specifically, one or more convolution scores may be generated between a prior map and a current map. Convolution scores may be generated by applying a fast Fourier transform on both the prior and current maps, multiplying the results of the transforms, and taking the inverse fast Fourier transform of the product. Based on these convolution scores, an autonomous vehicle may determine the offset between the maps and localize itself relative to the prior map.
1. A method for controlling a vehicle, the method comprising:
determining a current location of the vehicle;
receiving prior map information associated with the current location of the vehicle;
receiving current map information associated with the current location of the vehicle;
generating, using a processor, a score image based on the prior map information and the current map information, wherein the score image is generated using a fast Fourier transform;
identifying a plurality of offsets between the prior map information and current map information;
determining, using the score image, whether a given one of the plurality of offsets represents a shift between the prior map information and the current map information; and
correcting the position of the vehicle associated with the prior map information based on the given offset.
2. The method of claim 1 , wherein determining the current location of the vehicle comprises:
receiving data from a geographic position component;
identifying map information based on the received data; and
correcting the current location of the vehicle based on the identified map information.
3. The method of claim 1 , wherein the current map information is based on sensor information received by the vehicle.
4. The method of claim 3 wherein the current map information is image based.
5. The method of claim 1 , wherein the prior map information is image based.
6. The method of claim 1 , wherein the score image is represented by a plurality of pixels, and wherein the pixels correspond to convolution scores.
7. The method of claim 6 , wherein the plurality of offsets represents one or more of the plurality of pixels.
8. The method of claim 7 , wherein one or more of the plurality of offsets is associated with a horizontal component and a vertical component.
9. The method of claim 8 , wherein a given convolution score represents the convolution score between the prior map information and current map information at a given offset.
10. The method of claim 9 , wherein the given one of the plurality of offsets is the offset associated with the one or more pixels corresponding to the largest convolution score.
11. A method for controlling a vehicle, the method comprising:
determining a current location of the vehicle;
receiving prior map information associated with the current location of the vehicle;
receiving current map information associated with the current location of the vehicle;
identifying a plurality of offsets between the prior map information and current map information;
generating, using a processor, a plurality of convolution scores based on the plurality of offsets;
determining, based on the plurality of convolution scores whether a given one of the plurality of offsets represents a shift between the prior map information and current map information; and
determining a position of the vehicle within the prior map information based on the given offset.
12. The method of claim 11 , wherein determining the current location of the vehicle comprises:
receiving data from a geographic position component;
identifying map information based on the received data; and
correcting the current location of the vehicle based on the identified map information.
13. The method of claim 11 , wherein the current map information is based on sensor information received by the vehicle.
14. The method of claim 13 wherein the current map information is image based.
15. The method of claim 11 , wherein the prior map information is image based.
16. The method of claim 11 , wherein a convolution score is determined for each of the identified offsets.
17. The method of claim 16 , wherein the given one of the plurality of offsets is the identified offset associated with the largest convolution score.
18. An apparatus comprising:
a processor in communication with a memory and configured to execute instructions stored in the memory to determine a current location of the vehicle;
the memory storing instructions and map information associated with the current location of the vehicle;
the processor configured to execute the instructions stored in the memory to:
receive prior map information associated with the current location of the vehicle;
receive current map information associated with the current location of the vehicle;
generate a score image based on the prior map information and the current map information, wherein the score image is generated using a fast Fourier transform;
identify a plurality of offsets between the prior map information and current map information;
determine, using the score image, whether a given one of the plurality of offsets represents a shift between the prior map information and the current map information; and
determine a position of the vehicle associated with the prior map information based on the given offset.
19. An apparatus comprising:
a processor in communication with a memory and configured to execute instructions stored in the memory to determine a current location of the vehicle;
the memory storing instructions and map information associated with the current location of the vehicle;
the processor configured to execute the instructions stored in the memory to:
receive prior map information associated with the current location of the vehicle;
receive current map information associated with the current location of the vehicle;
identify a plurality of offsets between the prior map information and current map information;
generate a plurality of convolution scores based on the plurality of offsets;
determine, based on the plurality of convolution scores, whether a given one of the plurality of offsets represents a shift between the prior map information and current map information; and
determine a position of the vehicle within the prior map information based on the given offset.