Drive-through calibration process
System, methods, and computer-readable media for a calibration check process of sensors on an autonomous vehicle (AV) via a drive-through environment mapped with a series of targets. The targets are used to check the calibration of the sensors on the AV as a quality check before the AV is determined fit to drive autonomously. Supervising a calibration process that collects data from the sensors on the AV based on exposure to a series of targets or augmented camera targets. The collected data may be used for extrinsic calibration aimed to determine that extrinsic parameters that define the rigid relationship between sensors are in set checker bounds.
1. A method comprising:
initiating a calibration check process of an autonomous vehicle (AV) including a plurality of calibration checks, wherein during the calibration check process the AV drives along a path in an environment that provides exposure to a series of targets;
receiving, by a computing device of the AV, a set of data from a plurality of sensors on the AV to determine a status of the plurality of calibration checks, implemented by the computing device of the AV, including an odometry checker, a chassis motion Light Detection and Ranging (LiDAR) checker, a relative LiDAR checker, a camera extrinsics checker, a camera signals checker, and a Radio Detection and Ranging (RADAR) checker;
initiating the odometry checker in a turning segment of the path of the calibration check process;
in response to the initiation, comparing, by the odometry checker, an odometry trajectory to one or more reference poses of an inertial navigation system (INS); and
based on the comparison, detecting, by the odometry checker, deviations, including spatial drift and yaw drift, by the odometry trajectory and determining whether the deviations are with a determine bound;
initiating, by the computing device of the AV, the chassis motion LiDAR checker in a straight segment of the calibration check process;
determining, by the chassis motion LIDAR checker, chassis motion based on the odometry trajectory;
receiving, by the chassis motion LIDAR checker, real-time LiDAR data from a reference LiDAR sensor on the AV traveling within a specific speed limit;
comparing, by the chassis motion LIDAR checker, the real-time LiDAR data to the chassis motion for determining LiDAR-to-chassis extrinsics of the reference LiDAR sensor;
initiating, by the computing device of the AV, the relative LiDAR checker of the calibration check process;
determining, by the relative LiDAR checker, relative alignment of a plurality of LiDAR sensors with the reference LiDAR sensor;
initiating, by the computing device of the AV, the camera extrinsics checker in the calibration check process;
evaluating, by the camera extrinsics checker, extrinsics, including rotation and translation, of a plurality of cameras on the AV relative to a front central camera;
initiating, by the computing device of the AV, the camera signals checker of signal cameras in the calibration check process, wherein the signal cameras comprise a hardware filter for viewing traffic light bulbs;
detecting, by the camera signals checker, a dual traffic light target set a plurality of times during a straight drive portion of the calibration check process;
computing, by the camera signals checker, homography between dual traffic light targets each of the times, wherein a homography matrix is established to check relative pose error;
initiating, by the computing device of the AV, the RADAR checker in the calibration check process;
checking, by the RADAR checker, alignment of a plurality of RADAR sensors by comparing locations of detected target locations in the RADAR sensors against target locations detected by a LiDAR sensor on the AV; and
determining the status of the calibration checks that indicates whether the AV failed any individual sensor checkers or the calibration check process as a whole.
2. The method of claim 1 , wherein the path includes at least one turn and at least one straight path exceeding a minimum distance.
3. The method of claim 1 , where the calibration check process is initiated via remote operator, wherein the method further comprising:
providing an interface for the remote operator, wherein the interface includes a start calibration button, a stop calibration button, and indications of whether the AV passed or failed any of the plurality of calibration checks and/or a breakdown of whether each sensor passed or failed each of the calibration checks.
4. A non-transitory computer-readable medium comprising instructions stored thereon, when executed the instructions are effective to cause a system to:
initiate a calibration check process of an autonomous vehicle (AV) including a plurality of calibration checks, wherein during the calibration check process the AV drives along a path in an environment that provides exposure to a series of targets;
receive, by a computing device of the AV, a set of data from a plurality of sensors on the AV to determine a status of the plurality of calibration checks, implemented by the computing device of the AV, including an odometry checker, a chassis motion Light Detection and Ranging (LiDAR) checker, a relative LiDAR checker, a camera extrinsics checker, a camera signals checker, and a Radio Detection and Ranging (RADAR) checker;
initiate the odometry checker in a turning segment of the path of the calibration check process;
in response to the initiation, compare, by the odometry checker, an odometry trajectory to one or more reference poses of an inertial navigation system (INS); and
based on the comparison, detect, by the odometry checker, deviations, including spatial drift and yaw drift, by the odometry trajectory and determining whether the deviations are with a determine bound;
initiate, by the computing device of the AV, the chassis motion LiDAR checker in a straight segment of the calibration check process;
determine, by the chassis motion LIDAR checker, chassis motion based on the odometry trajectory;
receive, by the chassis motion LIDAR checker, real-time LiDAR data from a reference LiDAR sensor on the AV traveling within a specific speed limit;
compare, by the chassis motion LIDAR checker, the real-time LiDAR data to the chassis motion for determining LiDAR-to-chassis extrinsics of the reference LiDAR sensor;
initiate, by the computing device of the AV, the relative LiDAR checker of the calibration check process;
determine, by the relative LiDAR checker, relative alignment of a plurality of LiDAR sensors with the reference LiDAR sensor;
initiate, by the computing device of the AV, the camera extrinsics checker in the calibration check process;
evaluate, by the camera extrinsics checker, extrinsics, including rotation and translation, of a plurality of cameras on the AV relative to a front central camera;
initiate, by the computing device of the AV, the camera signals checker of signal cameras in the calibration check process, wherein the signal cameras comprise a hardware filter for viewing traffic light bulbs;
detect, by the camera signals checker, a dual traffic light target set a plurality of times during a straight drive portion of the calibration check process;
compute, by the camera signals checker, homography between dual traffic light targets each of the times, wherein a homography matrix is established to check relative pose error;
initiate, by the computing device of the AV, the RADAR checker in the calibration check process;
check, by the RADAR checker, alignment of a plurality of RADAR sensors by comparing locations of detected target locations in the RADAR sensors against target locations detected by a LiDAR sensor on the AV; and
determine the status of the calibration checks that indicates whether the AV failed any individual sensor checkers or the calibration check process as a whole.
5. The non-transitory computer-readable medium of claim 4 , wherein the path includes at least one turn and at least one straight path exceeding a minimum distance.
6. The non-transitory computer-readable medium of claim 4 , where the calibration check process is initiated via remote operator, wherein the instructions are effective to further cause the system to:
provide an interface for the remote operator, wherein the interface includes a start calibration button, a stop calibration button, and indications of whether the AV passed or failed any of the plurality of calibration checks and/or a breakdown of whether each sensor passed or failed each of the calibration checks.
7. A system comprising:
at least one processor; and
a memory storing computer-readable instructions that, when executed by the at least one processor, cause the system to:
initiate a calibration check process of an autonomous vehicle (AV) including a plurality of calibration checks, wherein during the calibration check process the AV drives along a path in an environment that provides exposure to a series of targets;
receive, by a computing device of the AV, a set of data from a plurality of sensors on the AV to determine a status of the plurality of calibration checks, implemented by the computing device of the AV, including an odometry checker, a chassis motion Light Detection and Ranging (LiDAR) checker, a relative LiDAR checker, a camera extrinsics checker, a camera signals checker, and a Radio Detection and Ranging (RADAR) checker;
initiate the odometry checker in a turning segment of the path of the calibration check process;
in response to the initiation, compare, by the odometry checker, an odometry trajectory to one or more reference poses of an inertial navigation system (INS); and
based on the comparison, detect, by the odometry checker, deviations, including spatial drift and yaw drift, by the odometry trajectory and determining whether the deviations are with a determine bound;
initiate, by the computing device of the AV, the chassis motion LiDAR checker in a straight segment of the calibration check process;
determine, by the chassis motion LIDAR checker, chassis motion based on the odometry trajectory;
receive, by the chassis motion LIDAR checker, real-time LiDAR data from a reference LiDAR sensor on the AV traveling within a specific speed limit;
compare, by the chassis motion LIDAR checker, the real-time LiDAR data to the chassis motion for determining LiDAR-to-chassis extrinsics of the reference LiDAR sensor;
initiate, by the computing device of the AV, the relative LiDAR checker of the calibration check process;
determine, by the relative LiDAR checker, relative alignment of a plurality of LiDAR sensors with the reference LiDAR sensor;
initiate, by the computing device of the AV, the camera extrinsics checker in the calibration check process;
evaluate, by the camera extrinsics checker, extrinsics, including rotation and translation, of a plurality of cameras on the AV relative to a front central camera;
initiate, by the computing device of the AV, the camera signals checker of signal cameras in the calibration check process, wherein the signal cameras comprise a hardware filter for viewing traffic light bulbs;
detect, by the camera signals checker, a dual traffic light target set a plurality of times during a straight drive portion of the calibration check process;
compute, by the camera signals checker, homography between dual traffic light targets each of the times, wherein a homography matrix is established to check relative pose error;
initiate, by the computing device of the AV, the RADAR checker in the calibration check process;
check, by the RADAR checker, alignment of a plurality of RADAR sensors by comparing locations of detected target locations in the RADAR sensors against target locations detected by a LiDAR sensor on the AV; and
determine, by the computing device of the AV, the status of the calibration checks that indicates whether the AV failed any individual sensor checkers or the calibration check process as a whole.
8. The system of claim 7 , wherein the path includes at least one turn and at least one straight path exceeding a minimum distance.
9. The system of claim 7 , where the calibration check process is initiated via remote operator, wherein the computer-readable instructions that, when executed by the at least one processor, further cause the system to:
provide an interface for the remote operator, wherein the interface includes a start calibration button, a stop calibration button, and indications of whether the AV passed or failed any of the plurality of calibration checks and/or a breakdown of whether each sensor passed or failed each of the calibration checks.