Methods and apparatus for calibrating stereo imagery using motion of vehicle
A system includes sensors onboard an autonomous vehicle, a processor, and a memory. The memory stores instructions for the processor to receive a first image pair from the sensors at a first time and a second image pair from the sensors at a second time. Each image pair includes at least one static feature in an environment of the autonomous vehicle. The memory also stores instructions to determine a distance travelled by the autonomous vehicle between the first and second times, and to determine a correction to a disparity map based on (1) a first disparity associated with the static feature(s) and the first image pair, (2) a second disparity associated with the static feature(s) and the second image pair, and (3) the distance travelled. The memory also stores instructions to cause the correction to be applied to the disparity map.
1 . A computer-implemented method comprising:
determining, by a computing system, a disparity correction for calibration of image data based on image pairs of a feature in an environment;
determining, by the computing system, a failure case associated with reduction in depth accuracy based on the disparity correction; and
updating, by the computing system, a map of the environment to identify the feature as a cause of the failure case.
2 . The computer-implemented method of claim 1 , further comprising:
determining, by the computing system, at least one image pair of the image pairs fails to satisfy a threshold number of correction datapoints; and
ignoring, by the computing system, the at least one image pair of the image pairs based on the threshold number of correction datapoints.
3 . The computer-implemented method of claim 1 , further comprising:
determining, by the computing system, a sparsity of a disparity map associated with the image data based on a lack of static objects depicted in the image data that satisfy a threshold distance from each other, wherein the failure case is determined based on the sparsity of the disparity map.
4 . The computer-implemented method of claim 1 , further comprising:
determining, by the computing system, an object depicted in the image data associated with the failure case; and
determining, by the computing system, the object is out of a field of view based on at least one of: an optical flow algorithm and a tracking algorithm.
5 . The computer-implemented method of claim 1 , wherein the updating the map of the environment comprises:
labeling, by the computing system, a structure in the environment as at least one of: repeated and misleading based on the failure case.
6 . The computer-implemented method of claim 1 , wherein the determining the failure case is based on identification of repeated structures in the image pairs.
7 . The computer-implemented method of claim 1 , wherein the determining the failure case is based on inaccuracies in reprojection of the image pairs.
8 . The computer-implemented method of claim 1 , further comprising:
determining, by the computing system, movement of one or more cameras associated with the image data; and
filtering, by the computing system, disparity corrections based on the movement of the one or more cameras.
9 . The computer-implemented method of claim 1 , further comprising:
determining, by the computing system, a frame of the image data exceeds a threshold value for variance; and
ignoring, by the computing system, the frame of the image data based on the threshold value for variance.
10 . The computer-implemented method of claim 1 , further comprising:
filtering, by the computing system, disparity corrections to a disparity map generated from the image data based on the failure case.
11 . A system comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the system to perform operations comprising:
determining a disparity correction for calibration of image data based on image pairs of a feature in an environment;
determining a failure case associated with reduction in depth accuracy based on the disparity correction; and
updating a map of the environment to identify the feature as a cause of the failure case.
12 . The system of claim 11 , the operations further comprising:
determining at least one image pair of the image pairs fails to satisfy a threshold number of correction datapoints; and
ignoring the at least one image pair of the image pairs based on the threshold number of correction datapoints.
13 . The system of claim 11 , the operations further comprising:
determining a sparsity of a disparity map associated with the image data based on a lack of static objects depicted in the image data that satisfy a threshold distance from each other, wherein the failure case is determined based on the sparsity of the disparity map.
14 . The system of claim 11 , the operations further comprising:
determining an object depicted in the image data associated with the failure case; and
determining the object is out of a field of view based on at least one of: an optical flow algorithm and a tracking algorithm.
15 . The system of claim 11 , wherein the updating the map of the environment comprises:
labeling a structure in the environment as at least one of: repeated and misleading based on the failure case.
16 . A non-transitory computer-readable storage medium including instructions that, when executed by at least on processor of a computing system, cause the computing system to perform operations comprising:
determining a disparity correction for calibration of image data based on image pairs of a feature in an environment;
determining a failure case associated with reduction in depth accuracy based on the disparity correction; and
updating a map of the environment to identify the feature as a cause of the failure case.
17 . The non-transitory computer-readable storage medium of claim 16 , the operations further comprising:
determining at least one image pair of the image pairs fails to satisfy a threshold number of correction datapoints; and
ignoring the at least one image pair of the image pairs based on the threshold number of correction datapoints.
18 . The non-transitory computer-readable storage medium of claim 16 , the operations further comprising:
determining a sparsity of a disparity map associated with the image data based on a lack of static objects depicted in the image data that satisfy a threshold distance from each other, wherein the failure case is determined based on the sparsity of the disparity map.
19 . The non-transitory computer-readable storage medium of claim 16 , the operations further comprising:
determining an object depicted in the image data associated with the failure case; and
determining the object is out of a field of view based on at least one of: an optical flow algorithm and a tracking algorithm.
20 . The non-transitory computer-readable storage medium of claim 16 , wherein the updating the map of the environment comprises:
labeling a structure in the environment as at least one of: repeated and misleading based on the failure case.