Performing 3D reconstruction via an unmanned aerial vehicle
In some examples, an unmanned aerial vehicle (UAV) employs one or more image sensors to capture images of a scan target and may use distance information from the images for determining respective locations in three-dimensional (3D) space of a plurality of points of a 3D model representative of a surface of the scan target. The UAV may compare a first image with a second image to determine a difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV. Further, based at least on the difference, the UAV may determine, while the UAV is in flight, an update to the 3D model including at least one of an updated location of at least one point in the 3D model, or a location of a new point in the 3D model.
1. An unmanned aerial vehicle (UAV) comprising:
one or more image sensors;
a propulsion mechanism; and
one or more processors configured by executable instructions to perform operations comprising:
capturing, with the one or more image sensors, while the UAV is in flight, a plurality of images of a scan target;
using distance information determined based on the plurality of images for determining respective locations in three-dimensional (3D) space of a plurality of points of a 3D model, the plurality of points representative of a surface of the scan target;
comparing a first image of the plurality of images with a second image of the plurality of images to determine one or more correspondences between the first image and the second image;
based at least in part on the one or more correspondences, determining, by the one or more processors, a difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV; and
based on at least one of the difference or the one or more correspondences, determining, by the one or more processors, while the UAV is in flight, an update to the 3D model including at least one of:
an updated location of at least one point in the 3D model, or
a location of a new point in the 3D model.
2. The UAV as recited in claim 1 , wherein the operation of determining, based at least on the difference, an update to the 3D model comprises:
receiving additional distance information based on additional captured images; and
determining the location of the new point in the 3D model based at least on the difference determined between the current frame of reference position for the UAV and the estimate of the actual frame of reference position for the UAV.
3. The UAV as recited in claim 1 , the operations further comprising:
based at least in part on the update to the 3D model, updating, while the UAV is in flight, a scan plan for scanning at least a portion of the scan target, the scan plan including a plurality of poses for the UAV to assume to capture images of the scan target, wherein a respective pose of the scan plan is determined based on a respective point of the 3D model; and
controlling a propulsion mechanism of the UAV to cause the UAV to fly to at least one of an additional pose or an updated pose included in the updated scan plan.
4. The UAV as recited in claim 1 , wherein the operation of using distance information determined based on the plurality of images for determining respective locations in 3D space of the plurality of points of the 3D model further comprises performing truncated signed distance function (TSDF) fusion using at least two images of the plurality of images of the scan target.
5. The UAV as recited in claim 1 , the operations further comprising:
grouping together a plurality of pairs of images of the plurality of images, including the first image and the second image; and
updating relative locations of multiple points of the 3D model based on comparing respective locations of respective features in the multiple pairs of images.
6. The UAV as recited in claim 5 , the operations further comprising:
determining correspondences between a plurality of higher-resolution images based at least on a plurality of the differences between the current frame of reference positions for the UAV and estimated actual frame of reference positions for the UAV for the plurality of higher-resolution images; and
generating, based at least in part on the determined correspondences, a higher-resolution 3D model that is of a higher resolution than the 3D model.
7. A method comprising:
capturing, by one or more processors of an unmanned aerial vehicle (UAV), with one or more image sensors, while the UAV is in flight, a plurality of images of a scan target;
using, by the one or more processors, distance information determined based on the plurality of images for determining respective locations in three-dimensional (3D) space of a plurality of points of a 3D model, the plurality of points representative of a surface of the scan target;
comparing, by the one or more processors, a first image of the plurality of images with a second image of the plurality of images to determine a difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV; and
based at least on the difference, determining, by the one or more processors, while the UAV is in flight, an update to the 3D model including at least one of:
an updated location of at least one point in the 3D model, or
a location of a new point in the 3D model.
8. The method as recited in claim 7 , wherein determining, based at least on the difference, an update to the 3D model comprises:
receiving additional distance information based on additional captured images; and
determining the location of the new point in the 3D model based at least on the difference determined between the current frame of reference position for the UAV and the estimate of the actual frame of reference position for the UAV.
9. The method as recited in claim 7 , further comprising:
based at least in part on the update to the 3D model, updating, by the one or more processors, while the UAV is in flight, a scan plan for scanning at least a portion of the scan target, the scan plan including a plurality of poses for the UAV to assume to capture images of the scan target, wherein a respective pose of the scan plan is determined based on a respective point of the 3D model; and
controlling, by the one or more processors, a propulsion mechanism of the UAV to cause the UAV to fly to at least one of an additional pose or an updated pose included in the updated scan plan.
10. The method as recited in claim 7 , wherein using distance information determined based on the plurality of images for determining respective locations in 3D space of the plurality of points of the 3D model further comprises performing truncated signed distance function (TSDF) fusion using at least two images of the plurality of images of the scan target.
11. The method as recited in claim 7 , wherein comparing the first image of the plurality of images with the second image of the plurality of images to determine the difference further comprises:
comparing the first image with the second image to determine one or more correspondences between the first image and the second image; and
based at least in part on the one or more correspondences, determining, by the one or more processors, the difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV.
12. The method as recited in claim 7 , further comprising:
grouping together a plurality of pairs of images of the plurality of images, including the first image and the second image; and
updating relative locations of multiple points of the 3D model based on comparing respective locations of respective features in the multiple pairs of images.
13. The method as recited in claim 7 , further comprising:
determining correspondences between a plurality of higher-resolution images based at least on a plurality of the differences between the current frame of reference positions for the UAV and estimated actual frame of reference positions for the UAV for the plurality of higher-resolution images; and
generating, based at least in part on the determined correspondences, a higher-resolution 3D model that is of a higher resolution than the 3D model.
14. An unmanned aerial vehicle (UAV) comprising
one or more image sensors;
a propulsion mechanism; and
one or more processors configured by executable instructions to perform operations comprising:
capturing, with the one or more image sensors, while the UAV is in flight, a plurality of images of a scan target;
using distance information determined based on the plurality of images for determining respective locations in three-dimensional (3D) space of a plurality of points of a 3D model, the plurality of points representative of a surface of the scan target;
comparing a first image of the plurality of images with a second image of the plurality of images to determine a difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV; and
based at least on the difference, determining, while the UAV is in flight, an update to the 3D model including at least one of:
an updated location of at least one point in the 3D model, or
a location of a new point in the 3D model.
15. The UAV as recited in claim 14 , wherein the operation of determining, based at least on the difference, an update to the 3D model comprises:
receiving additional distance information based on additional captured images; and
determining the location of the new point in the 3D model based at least on the difference determined between the current frame of reference position for the UAV and the estimate of the actual frame of reference position for the UAV.
16. The UAV as recited in claim 14 , the operations further comprising:
based at least in part on the update to the 3D model, updating, while the UAV is in flight, a scan plan for scanning at least a portion of the scan target, the scan plan including a plurality of poses for the UAV to assume to capture images of the scan target, wherein a respective pose of the scan plan is determined based on a respective point of the 3D model; and
controlling a propulsion mechanism of the UAV to cause the UAV to fly to at least one of an additional pose or an updated pose included in the updated scan plan.
17. The UAV as recited in claim 14 , wherein the operation of using distance information determined based on the plurality of images for determining respective locations in 3D space of the plurality of points of the 3D model further comprises performing truncated signed distance function (TSDF) fusion using at least two images of the plurality of images of the scan target.
18. The UAV as recited in claim 14 , wherein the operation of comparing the first image of the plurality of images with the second image of the plurality of images to determine the difference further comprises:
comparing the first image with the second image to determine one or more correspondences between the first image and the second image; and
based at least in part on the one or more correspondences, determining, by the one or more processors, the difference between a current frame of reference position for the UAV and an estimate of an actual frame of reference position for the UAV.
19. The UAV as recited in claim 14 , the operations further comprising:
grouping together a plurality of pairs of images of the plurality of images, including the first image and the second image; and
updating relative locations of multiple points of the 3D model based on comparing respective locations of respective features in the multiple pairs of images.
20. The UAV as recited in claim 14 , the operations further comprising:
determining correspondences between a plurality of higher-resolution images based at least on a plurality of the differences between the current frame of reference positions for the UAV and estimated actual frame of reference positions for the UAV for the plurality of higher-resolution images; and
generating, based at least in part on the determined correspondences, a higher-resolution 3D model that is of a higher resolution than the 3D model.