Image space motion planning of an autonomous vehicle
An autonomous vehicle that is equipped with image capture devices can use information gathered from the image capture devices to plan a future three-dimensional (3D) trajectory through a physical environment. To this end, a technique is described for image-space based motion planning. In an embodiment, a planned 3D trajectory is projected into an image-space of an image captured by the autonomous vehicle. The planned 3D trajectory is then optimized according to a cost function derived from information (e.g., depth estimates) in the captured image. The cost function associates higher cost values with identified regions of the captured image that are associated with areas of the physical environment into which travel is risky or otherwise undesirable. The autonomous vehicle is thereby encouraged to avoid these areas while satisfying other motion planning objectives.
1. A method comprising:
generating a cost function map that associates a cost value with each of multiple regions of an image of a physical environment, wherein the cost value is indicative of a level of risk associated with navigating in an area of the physical environment corresponding to a particular region of the image; and
causing a vehicle to autonomously maneuver through the physical environment using the cost function map.
2. The method of claim 1 , wherein the cost value of the particular region is based on a determined level of confidence in an estimated depth to a physical surface in an area of the physical environment corresponding to the particular region of the image.
3. The method of claim 1 , wherein the cost value of the particular region is based on a detected physical object in an area of the physical environment corresponding to the particular region of the image.
4. The method of claim 1 , wherein generating the cost function map includes:
inputting the image into a machine learning model to determine the level of risk associated with navigating in the area of the physical environment corresponding to the particular region of the image.
5. The method of claim 1 , wherein causing the vehicle to autonomously maneuver through the physical environment using the cost function map includes:
adjusting a planned 3D trajectory for the vehicle through the physical environment based on a relationship between the cost function map and a projection of the planned 3D trajectory.
6. The method of claim 1 , wherein causing the vehicle to autonomously maneuver includes:
generating control commands configured to cause the vehicle to autonomously maneuver.
7. The method of claim 1 , further comprising:
processing additional images to continually update the cost function map; and
causing the vehicle to autonomously maneuver through the physical environment using the updated cost function map.
8. The method of claim 1 , wherein the image of the physical environment is captured by a camera coupled to the vehicle.
9. The method of claim 1 , wherein the vehicle is an unmanned aerial vehicle (UAV).
10. A vehicle comprising:
an image sensor configured to capture images of a physical environment;
a propulsion system configured to maneuver the vehicle through the physical environment; and
a visual navigation system coupled with the image sensor and the propulsion system, the visual navigation system configured to:
generate a cost function map,
wherein the cost function map associates a cost value with each of multiple regions of an image of the physical environment,
wherein the cost value is indicative of a level of risk associated with navigating in an area of the physical environment corresponding to a particular region of the image; and
direct the propulsion system to cause the vehicle to autonomously maneuver through the physical environment using the cost function map.
11. The autonomous vehicle of claim 10 , wherein the cost value of the particular region is based on a determined level of confidence in an estimated depth to a physical surface in an area of the physical environment corresponding to the particular region of the image.
12. The autonomous vehicle of claim 10 , wherein the cost value of the particular region is based on a detected physical object in an area of the physical environment corresponding to the particular region of the image.
13. The autonomous vehicle of claim 10 , wherein to generate the cost function map, the visual navigation system is configured to:
input the image into a machine learning model to determine the level of risk associated with navigating in the area of the physical environment corresponding to the particular region of the image.
14. The autonomous vehicle of claim 10 , wherein to direct the propulsion system to cause the vehicle to autonomously maneuver through the physical environment using the cost function map, the visual navigation system is configured to:
adjust a planned 3D trajectory for the vehicle through the physical environment based on a relationship between the cost function map and a projection of the planned 3D trajectory.
15. The autonomous vehicle of claim 10 , wherein the visual navigation system is configured to:
process additional images to continually update the cost function map; and
direct the propulsion system to cause the vehicle to autonomously maneuver through the physical environment using updated cost function map.
16. An apparatus, comprising:
one or more memory units storing instructions that, when executed by one or more processors, cause the one or more processors to:
generate a cost function map, the cost function map associating a cost value with each of multiple regions of an image, wherein the cost value is indicative of a level of risk associated with navigating in an area of the physical environment corresponding to a particular region of the image; and
cause a vehicle to autonomously maneuver through the physical environment using the cost function map.
17. The apparatus of claim 16 , wherein the cost value of the particular region is based on one or more of:
a determined level of confidence in an estimated depth to a physical surface in an area of the physical environment corresponding to the particular region of the image; and
a detected physical object in an area of the physical environment corresponding to the particular region of the image.
18. The apparatus of claim 16 , wherein to generate the cost function map, the instructions, when executed by the one or more processors, cause the one or more processors to:
input the image into a machine learning model to determine the level of risk associated with navigating in the area of the physical environment corresponding to the particular region of the image.
19. The apparatus of claim 16 , wherein to cause the vehicle to autonomously maneuver through the physical environment using the cost function map, the instructions, when executed by the one or more processors, cause the one or more processors to:
adjust a planned 3D trajectory for the vehicle through the physical environment based on a relationship between the cost function map and a projection of the planned 3D trajectory.
20. The apparatus of claim 16 , wherein the instructions, when executed by the one or more processors, cause the one or more processors to:
process additional images of the physical environment to continually update the cost function map while the vehicle is in motion through the physical environment; and
cause a vehicle to autonomously maneuver through the physical environment based on the continually updated cost function map.