Table 10 Variance of estimated attitude error in real world (#4)
From: EKF-based self-attitude estimation with DNN learning landscape information
| Â | Roll [\({\text{deg}}^2\)] | Pitch [\({\text{deg}}^2\)] |
|---|---|---|
MLE (ours, all) | 20.548 | 20.585 |
MLE (ours, selected) | 9.356 | 13.971 |
Regression w/ L2 normalization | 20.976 | 20.982 |
Regression w/o L2 normalization | 25.996 | 24.160 |