GNSS Calibration Position Uncertainty

Software Python
1

I am attempting to acheive GNSS calibration with data stored in svo2 data keys.
During the svo processing steps, I get my calibration parameters with the following method:

    def fusion_status(self) -> None:
        process_metrics: Tuple[sl.FUSION_ERROR_CODE,sl.FusionMetrics] = self.fusion.get_process_metrics()
        fusion_error_code, fusion_metrics = process_metrics
        fusion_metrics = {"mean_camera_fused":fusion_metrics.mean_camera_fused,"mean_stdev_camera_ts":fusion_metrics.mean_stdev_between_camera}


        vio_gnss_trns: sl.Transform = self.fusion.get_geo_tracking_calibration()
        vio_gnss_trns_translation:np.ndarray = vio_gnss_trns.get_translation().get()

        gnss_calibration_data: Tuple[sl.GNSS_FUSION_STATUS,float,np.ndarray] = self.fusion.get_current_gnss_calibration_std()
        gnss_calibration, yaw_uncertainty, position_uncertainty = gnss_calibration_data

        if gnss_calibration.name == "OK":
            logger.info(f"Fusion: GNSS Calibration: {gnss_calibration.name}")
            # self.solve_local_cs()
    
        out = {
            "fusion_error_code": fusion_error_code.name,
            "fusion_metrics": fusion_metrics,
            "vio_gnss_trns": vio_gnss_trns_translation.tolist(),
            "gnss_calibration": gnss_calibration.name,
            "yaw_uncertainty": yaw_uncertainty,
            "position_uncertainty": position_uncertainty.tolist()

        }

These are the values i have in the debugger:

>>> out
{
'fusion_error_code': 'SUCCESS', 
'fusion_metrics': {
    'mean_camera_fused': 0.6666666865348816, 
    'mean_stdev_camera_ts': 0.0
    }, 'vio_gnss_trns': [
        -9.201568603515625,
        -8.376220703125,
        34.25571060180664
    ], 'gnss_calibration': 'OK', 
        'yaw_uncertainty': 0.001549591775983572, 
        'position_uncertainty': [
            
        1.8446742974197924e+19,
        0.8890160918235779,
        1.8446742974197924e+19
    ]
}

Any ideas why I am getting such large position uncertainties?

2

Hello @fdunbar,
Unfortunately, I cannot reproduce your issue. Could you share SVO (with GNSS data) + code sample for reproducing it ?
Regards

3

Before we get to that,
Is there a step I need to take where I “Initialize” my fusion with the first GNSS measurement available?

My visual odometry → GNSS translation is quite large:

vio_gnss_trns=[-29.21543312072754, -22.91114044189453, 34.25571060180664]