Performing low profile object detection on a mower

1. A mower comprising:

a main body;

one or more mower decks supported by the main body;

one or more sensors that provide sensor data for a detection area that is in a path of the one or more mower decks such that the detection area changes as the mower travels, the detection area corresponding to a field of view of the one or more sensors;

an autonomy controller that receives the sensor data for the detection area from the one or more sensors; and

a machine controller that controls a ground speed of the mower;

wherein the autonomy controller is configured to perform a method for detecting an object within the detection area, the method comprising:

while the mower travels at a first ground speed at a first particular time in an area to be mowed, processing the sensor data for the detection area that is in the path of the one or more mower decks at the first particular time to determine that an object may be present in a region of interest within the detection area that is in the path of the one or more mower decks at the first particular time, wherein the sensor data for the detection area that is in the path of the one or more mower decks at the first particular time corresponds to the field of view of the one or more sensors at the first particular time, and wherein the region of interest corresponds to a portion of the field of view of the one or more sensors at the first particular time;

in response to determining that an object may be present in the region of interest, causing the machine controller to slow the mower to a second ground speed to thereby slow a rate at which the one or more decks approaches the object that may be present in the region of interest;

while the mower continues to travel at the second ground speed in the area to be mowed and the region of interest remains in the detection area that is in the path of the one or more mower decks at the first particular time, buffering multiple samples of region of interest sensor data received over a period of time after the first particular time as the mower approaches the region of interest, the region of interest sensor data being a subset of the sensor data for the detection area that is in the path of the one or more mower decks during the period of time, the subset of the sensor data encompassing the region of interest such that the multiple samples of region of interest sensor data comprise increasingly closer views of the region of interest as the mower approaches the region of interest; and

while the mower continues to travel at the second ground speed in the area to be mowed and the region of interest remains in the detection area that is in the path of the one or more mower decks during the period of time, simultaneously processing the buffered multiple samples of region of interest sensor data which provide the increasingly closer views of the region of interest to determine whether an object is present in the region of interest.

2. The mower of claim 1 , wherein the one or more sensors comprise one or more 2D sensors and one or more 3D sensors.

3. The mower of claim 2 , wherein processing the sensor data for the detection area that is in the path of the one or more mower decks comprises processing sensor data for the detection area received from the one or more 2D sensors.

4. The mower of claim 3 , wherein the autonomy controller receives the region of interest sensor data from the one or more 3D sensors.

5. The mower of claim 1 , wherein, in response to determining that an object may be present in the region of interest within the detection area that is in the path of the one or more mower decks, the autonomy controller specifies the region of interest to the one or more sensors to cause the one or more sensors to provide the region of interest sensor data over the period of time.

6. The mower of claim 1 , wherein, in response to determining that an object may be present in the region of interest within the detection area that is in the path of the one or more mower decks, the autonomy controller extracts the region of interest sensor data from the sensor data for the detection area that is in the path of the one or more mower decks received from the one or more sensors over the period of time.

7. The mower of claim 1 , wherein the method further comprises:

in response to determining that an object is present in the region of interest, causing the machine controller to alter a path of the mower.

8. The mower of claim 1 , wherein the method further comprises:

in response to determining that an object is not present in the region of interest, causing the machine controller to increase the ground speed of the mower.

9. The mower of claim 1 , wherein the method further comprises:

in response to failing to determine whether an object is present in the region of interest, notifying an external system.

10. The mower of claim 9 , wherein notifying the external system comprises providing a live video feed from a camera on the mower.

11. The mower of claim 1 , wherein processing the sensor data for the detection area that is in the path of the one or more mower decks comprises employing a first tolerance in a ground plane algorithm and wherein processing the buffered multiple samples of region of interest sensor data comprises employing a second tolerance in the ground plane algorithm, the second tolerance being tighter than the first tolerance.

12. The mower of claim 1 , wherein the autonomy controller continues to process the sensor data for the detection area that is in the path of the one or more mower decks while processing the buffered multiple samples of region of interest sensor data.

13. A method, performed by an autonomy controller of a vehicle, for detecting a low profile object, the method comprising:

receiving, from one or more sensors, sensor data for a detection area that is in a path of the vehicle at a first particular time such that the detection area changes as the vehicle travels, wherein the sensor data for the detection area that is in the path of the vehicle at the first particular time corresponds to a field of view of the one or more sensors at the first particular time;

processing the sensor data for the detection area that is in the path of the vehicle at the first particular time to determine that an object may be present in a region of interest within the detection area that is in the path of the vehicle at the first particular time, wherein the region of interest corresponds to a portion of the field of view of the one or more sensors at the first particular time;

in response to determining that an object may be present in the region of interest, causing a ground speed of the vehicle to be slowed to thereby slow a rate at which the vehicle approaches the object that may be present in the region of interest;

in conjunction with causing the ground speed of the vehicle to be slowed, buffering multiple samples of region of interest sensor data received over a period of time after the first particular time as the vehicle approaches the region of interest, the region of interest sensor data being a subset of the sensor data for the detection area that is in the path of the vehicle during the period of time, the subset of the sensor data encompassing the region of interest such that the multiple samples of region of interest sensor data comprise increasingly closer views of the region of interest as the vehicle approaches the region of interest; and

while the region of interest remains in the detection area that is in the path of the vehicle during the period of time, simultaneously processing the buffered multiple samples of region of interest sensor data which provide the increasingly closer views of the region of interest to determine whether an object is present in the region of interest.

14. The method of claim 13 , wherein the vehicle is a mower.

15. The method of claim 13 , wherein the sensor data for the detection area that is in the path of the vehicle is received from one or more 2D sensors.

16. The method of claim 15 , wherein the region of interest sensor data is received from one or more 3D sensors.

17. The method of claim 16 , wherein the sensor data for the detection area that is in the path of the vehicle is also received from the one or more 3D sensors.

18. The method of claim 13 , wherein:

in response to determining that an object may be present in the region of interest, the autonomy controller specifies the region of interest to the one or more sensors to cause the one or more sensors to provide the region of interest sensor data over the period of time; or

in response to determining that an object may be present in the region of interest, the autonomy controller extracts the region of interest sensor data from the sensor data for the detection area that is in the path of the vehicle received from the one or more sensors over the period of time.

19. The method of claim 13 , further comprising:

in response to failing to determine whether an object is present in the region of interest, notifying an external system.

20. The method of claim 19 , wherein notifying the external system comprises providing a feed from a camera of the vehicle to the external system, and wherein the method further comprises:

displaying, by the external system, the feed;

receiving, by the external system, user input that labels an object contained in the feed; and

causing the feed with the labeled object to be stored in a database.

21. A vehicle comprising:

a main body;

a first sensor and a second sensor that each provide sensor data for a detection area that is in a path of the main body such that the detection area changes as the vehicle travels;

an autonomy controller that receives the sensor data for the detection area that is in the path of the main body from the first and second sensors; and

a machine controller that controls a ground speed of the vehicle;

wherein the autonomy controller is configured to perform a method for detecting an object within the detection area that is in the path of the main body, the method comprising:

receiving, from one or both of the first and second sensor, sensor data for the detection area that is in the path of the main body at a first particular time, wherein the sensor data for the detection area that is in the path of the main body at the first particular time corresponds to a field of view of one or both of the first and second sensor at the first particular time;

processing the sensor data for the detection area that is in the path of the main body at the first particular time to determine that an object may be present in a region of interest within the detection area that is in the path of the main body, wherein the region of interest corresponds to a portion of the field of view of one or both of the first and second sensor at the first particular time;

in response to determining that an object may be present in the region of interest, causing the machine controller to slow the ground speed of the vehicle;

while the ground speed of the vehicle is slowed and while the region of interest remains in the detection area that is in the path of the main body at the first particular time, buffering multiple samples of region of interest sensor data received over a period of time after the first particular time from the second sensor, the region of interest sensor data being a subset of the sensor data for the detection area that is in the path of the main body during the period of time, the subset of the sensor data encompassing the region of interest such that the multiple samples of region of interest sensor data comprise increasingly closer views of the region of interest as the vehicle approaches the region of interest; and

while the region of interest remains in the detection area that is in the path of the main body during the period of time, simultaneously processing the buffered multiple samples of region of interest sensor data which provide the increasingly closer views of the region of interest to determine whether an object is present in the region of interest.

22. A method for building a database of labeled images for use in detecting a low profile object in a path of a vehicle, the method comprising:

receiving, from one or more sensors on a vehicle, sensor data for a detection area that is in a path of the vehicle at a first particular time such that the detection area changes as the vehicle travels, wherein the sensor data for the detection area that is in the path of the vehicle at the first particular time corresponds to a field of view of the one or more sensors at the first particular time;

processing the sensor data for the detection area that is in the path of the vehicle at the first particular time to determine that an object may be present in a region of interest within the detection area that is in the path of the vehicle at the first particular time, wherein the region of interest corresponds to a portion of the field of view of the one or more sensors at the first particular time;

in response to determining that an object may be present in the region of interest, causing a ground speed of the vehicle to be slowed to thereby slow a rate at which the vehicle approaches the object that may be present in the region of interest;

in conjunction with causing the ground speed of the vehicle to be slowed, buffering multiple samples of region of interest sensor data received over a period of time after the first particular time as the vehicle approaches the region of interest, the region of interest sensor data being a subset of the sensor data for the detection area that is in the path of the vehicle during the period of time, the subset of the sensor data encompassing the region of interest such that the multiple samples of region of interest sensor data comprise increasingly closer views of the region of interest as the vehicle approaches the region of interest;

while the region of interest remains in the detection area that is in the path of the vehicle during the period of time, simultaneously processing the buffered multiple samples of region of interest sensor data which provide the increasingly closer views of the region of interest to determine whether an object is present in the region of interest;

upon determining that an object may be present in the region of interest, providing a feed from a camera of the vehicle to an external system;

displaying, by the external system, the feed;

receiving, by the external system, user input that labels an object contained in the feed;

storing the feed with the labeled object in a database; and

using the feed with the labeled object to train and implement an artificial intelligence engine that is configured to detect objects in a path of the vehicle using a camera.

23. The method of claim 22 , wherein the vehicle is a mower.

24. The method of claim 22 , wherein the feed comprises one or more images.