1: \begin{abstract}

2:     % The abstract.  No math, special symbols, abbreviations, or citations.

3:     Many problems in robotics, such as estimating the state from noisy sensor data or aligning two LiDAR point clouds, can be posed and solved as least-squares problems.

4:     Unfortunately, vanilla nonminimal solvers for least-squares problems are notoriously sensitive to outliers.

5:     As such, various robust loss functions have been proposed to reduce the sensitivity to outliers.

6:     Examples of loss functions include pseudo-Huber, Cauchy, and Geman-McClure.

7:     Recently, these loss functions have been generalized into a single loss function that enables the best loss function to be found adaptively based on the distribution of the residuals.

8:     However, even with the generalized robust loss function, most nonminimal solvers can only be solved locally given a prior state estimate due to the nonconvexity of the problem.

9:     The first contribution of this paper is to combine graduated nonconvexity (GNC) with the generalized robust loss function to solve least-squares problems without a prior state estimate and without the need to specify a loss function.

10:     Moreover, existing loss functions, including the generalized loss function, are based on Gaussian-like distribution.

11:     However, residuals are often defined as the squared norm of a multivariate error and distributed in a Chi-like fashion.

12:     The second contribution of this paper is to apply a norm-aware adaptive robust loss function within a GNC framework.

13:     This leads to additional robustness when compared with state-of-the-art methods.

14:     Simulations and experiments demonstrate that the proposed approach is more robust and yields faster convergence times compared to other GNC formulations.

15: \end{abstract}

16: