Vehicle Self-Calibration

Operating Principle

NVIDIA^® DriveWorks uses vehicle model to calibrate the steering system properities using vehicle odometery and vehicle egomotion measurement and predicted vehicle motion based on vehicle model.

Steering Offset

Vehicle self-calibration supports vehicle steering offset self calibration. The steering offset is the difference between the actual steering and the measured steering angle, and it equals to the negative of the measured steering angle when the vehicle drive straight. The steering offset is calibrated by minimizing the predicted turn curvature based on model and curvature computed based on measurement data. To start the calibration process, the vehicle has to drive above 10 meter per seocnd, perform gradual lane change maneuvers.

Normalized steering offset q3 (steering offset / wheelbase) (center histogram), and Understeer coefficient (left histogram)collected over a period of time

Requirements

Initialization Requirements

• Nominal values on vehicle calibration
  • steering offset: 0 radian

Runtime Calibration Dependencies

• IMU-based egomotion needs to be based on accurate IMU calibration and odometry properties

Input Requirements

• Assumption: Vehicle performs the aforementioned maneuvers until calibration convergence.
• Vehicle egomotion: requirements can be found in the Egomotion module
• Vehicle IO State: requirements can be found in the VehicleIO VehicleIO

Output Requirements

• Corrected steering offset: less than 0.8 degrees accuracy

Cross-validation KPI

Several hours of data are used to produce a reference calibration value for cross-validation. Then, short periods of data are evaluated for whether they can recover the same values. For example, the graph below shows precision/recall curves of normalized steering offset q3 (q3=steering offset / wheelbase). Precision indicates that an accepted calibration is within a fixed precision threshold from the reference calibration, and recall indicates the ratio of accepted calibrations in the given amount of time.

Workflow

The following code snippet shows the general structure of a program that performs IMU self-calibration

dwCalibrationEngine_initialize(...);           // depends on sensor from rig configuration module
dwCalibrationEngine_initializeVehicle(...);    // depends on nominal calibration from rig configuration
dwCalibrationEngine_startCalibration(...);     // runtime calibration dependencies need to be met
 
while(true) // main loop
{
    // get current vehicle IO state
    dwVehicleIO_getVehicleNonSafetyState(&vehicleIONonSafetyState, ...);
    dwVehicleIO_getVehicleActuationFeedback(&vehicleIOActuationFeedback, ...);
 
    // feed vehicle io state into self-calibration
    dwCalibrationEngine_addVehicleIONonSafetyState(&vehicleIONonSafetyState, ...);
    dwCalibrationEngine_addVehicleIOActuationFeedback(&vehicleIOActuationFeedback, ...);
 
    // retrieve calibration status
    dwCalibrationEngine_getCalibrationStatus(...);
 
    // retrieve self-calibration results
    dwCalibrationEngine_getVehicleSteeringProperties(...);
}
 
dwCalibrationEngine_stopCalibration(...);