Systems and methods for blind online calibration of radar systems on a vehicle
Systems, methods, and non-transitory computer-readable media provide a blind online calibration mechanism to calibrate the position of the radar unit while the self-driving vehicle is in motion without the use of any map data. Specifically, a calibration system associated with the self-driving vehicle is configured to adjust the boresight angle of the radar within a calibration range and monitor the convergence or divergence pattern of the resulting clutter locations. The boresight angle of the radar unit may be progressively adjusted in static or dynamic degree increments until the convergence of the clutter curves is observed. In this way, without using any map data or factory settings, radar calibration can be conducted as often as needed while the vehicle is moving. Radar sensing and measurement accuracy is thus improved.
1. A method of calibrating a sensor unit on a vehicle while the vehicle is in motion, the method comprising:
transmitting, from a processor to the sensor unit, a control signal that causes a position of the sensor unit relative to an inertial measurement unit (IMU) to be tuned for positions between the IMU and the sensor unit while the vehicle is in motion;
obtaining sensor data scans captured by the sensor unit while the sensor unit is tuned to the positions;
identifying target points from the sensor data scans, wherein target points represent one or more static objects in an environment around the vehicle while the vehicle is in motion; and
determining a calibration position for the sensor unit based on a trajectory pattern of the target points over the sensor data scans.
2. The method of claim 1 , wherein the sensor unit is a radar unit, and the positions include an angle that the radar unit is facing relative to the IMU.
3. The method of claim 2 , wherein the sensor unit is tuned by sweeping the angle progressively at a pre-defined degree increment.
4. The method of claim 1 , wherein the sensor data scans are captured at a pre-defined interval, and at least one or more of the sensor data scans are captured at the positions to which the sensor unit is tuned.
5. The method of claim 1 , wherein the identifying the target points from of the sensor data scans represent one or more static objects in the environment located in while the vehicle is in motion comprises:
identifying a first set of target points from a first sensor data scan from the sensor data scans;
identifying a second set of target points from a second sensor data scan from the sensor data scans; and
in response to determining that an aggregated average distance between the first set of target points and the second set of target points within a sensor data scan frame is less than a pre-defined threshold, determining that the first set of target points and the second set of target points relate to a same static object in the environment.
6. The method of claim 1 , wherein determining the calibration position for the sensor unit based on the trajectory pattern of the target points over the sensor data scans comprises:
monitoring the trajectory pattern of the target points representing a same static object in the environment shown in the sensor data scans;
identifying a position for the sensor unit at which the target points representing the same static object shown in a corresponding sensor data scan captured at the position is associated with a minimum variance; and
determining the position as the calibration position for the sensor unit.
7. The method of claim 6 , wherein the position associated with the minimum variance is determined by:
obtaining a first sensor data scan captured when the sensor unit is tuned to a first position;
identifying a first set of target points from the first sensor data scan;
transmitting, to the sensor unit, a first control signal that causes the sensor unit to be progressively tuned in a first direction from the first position to a second position;
obtaining a second sensor data scan captured when the sensor unit is tuned to the second position;
identifying, from the second sensor data scan, a second set of target points representing the same static object with the first set of target points; and
in response to determining that the second set of target points is converging as compared with the first set of target points, transmitting, to the sensor unit, a second control signal that causes the sensor unit to be tuned in the first direction from the second position to a third position.
8. The method of claim 7 , further comprising:
in response to determining that the second set of target points is diverging as compared with the first set of target points, transmitting, to the sensor unit, a third control signal that causes the sensor unit to be tuned in a second direction that is opposite to the first direction.
9. The method of claim 1 , wherein the calibration position for the sensor unit is updated periodically or on demand while the vehicle is in motion.
10. The method of claim 1 , wherein the calibration position for the sensor unit is determined or updated adaptively using the sensor data scans captured in real: time without using map data indicating a location of the one or more static objects in the environment.
11. A system for calibrating a sensor unit on a vehicle while the vehicle is in motion, comprising:
one or more non-transitory computer-readable storage media embodying instructions; and
one or more processors coupled to the storage media and operable to execute the instructions to:
transmit, from a processor to the sensor unit, a control signal that causes the sensor unit installed with the vehicle to be tuned for a plurality of positions between an inertial measurement unit (IMU) and the sensor unit while the vehicle is in motion;
obtain a plurality of sensor data scans captured by the sensor unit while the sensor unit is tuned to the plurality of positions;
identify one or more target points from each of the plurality of sensor data scans, wherein the one or more target points represent one or more static objects in an environment that the vehicle is situated at while in motion; and
determine a calibration position for the sensor unit based on a trajectory pattern of the one or more target points over the plurality of sensor data scans.
12. The system of claim 11 , wherein the sensor unit is a radar unit, and each of the plurality of positions includes a boresight angle that the radar unit is facing, and
wherein the sensor unit is tuned by sweeping the boresight progressively at a pre-defined degree interval each time.
13. The system of claim 11 , wherein the plurality of sensor data scans are captured at a pre-defined interval, and at least one or more of the plurality of sensor data scans are captured at each of the positions to which the sensor unit is tuned.
14. The system of claim 11 , wherein identifying the one or more target points comprises:
identifying a first set of target points from a first sensor data scan from the plurality of sensor data scans;
identifying a second set of target points from a second sensor data scan from the plurality of sensor data scans; and
in response to determining that an aggregated average distance between the first set of target points and the second set of target points within a sensor data scan frame is less than a pre-defined threshold, determining that the first set of target points and the second set of target points relate to a same static object in the environment.
15. The system of claim 11 , wherein determining the calibration position comprises:
monitoring the trajectory pattern of the one or more target points representing a same static object in the environment shown in the plurality of sensor data scans;
identifying a position for the sensor unit at which the one or more target points representing the same static object shown in a corresponding sensor data scan captured at the position contain exhibit a minimum variance; and
determining the position as the calibration position for the sensor unit.
16. The system of claim 15 , wherein the position associated with the minimum variance is determined by:
obtaining a first sensor data scan captured when the sensor unit is tuned to a first position;
identifying a first set of target points from the first sensor data scan;
transmitting, to the sensor unit, a first control signal that causes the sensor unit to be progressively tuned in a first direction from the first position to a second position;
obtaining a second sensor data scan captured when the sensor unit is tuned to the second position;
identifying, from the second sensor data scan, a second set of target points representing the same static object with the first set of target points; and
in response to determining that the second set of target points is converging as compared with the first set of target points, transmitting, to the sensor unit, a second control signal that causes the sensor unit to be tuned in the first direction from the second position to a third position.
17. The system of claim 16 , wherein the one or more processors are further operable to execute the instructions to:
in response to determining that the second set of target points is diverging as compared with the first set of target points, transmit, to the sensor unit, a third control signal that causes the sensor unit to be tuned in a second direction that is different from the first direction.
18. The system of claim 11 , wherein the calibration position for the sensor unit is updated periodically or on demand while the vehicle is in motion.
19. The system of claim 11 , wherein the calibration position for the sensor unit is determined or updated adaptively using the plurality of sensor data scans captured in real time without using any map data indicating a location of the one or more static objects in the environment.
20. One or more non-transitory computer-readable storage media embodying software for calibrating a sensor unit on a vehicle while the vehicle is in motion and that is operable when executed by a computing system to:
transmit, from a processor to the sensor unit, a control signal that causes the sensor unit installed with the vehicle to be tuned for a plurality of positions while the vehicle is in motion;
obtain a plurality of sensor data scans captured by the sensor unit while the sensor unit is tuned to the plurality of positions;
identify one or more target points from each of the plurality of sensor data scans, wherein the one or more target points represent one or more static objects at a location that the vehicle is at while in motion; and
determine a calibration position for the sensor unit based on a trajectory pattern of the one or more target points over the plurality of sensor data scans.