1: \begin{abstract}

2: Semi-functional linear regression models postulate a linear

3: relationship between a scalar response and a functional covariate, and

4: also include a non-parametric component involving a univariate

5: explanatory variable. It is of practical importance to obtain

6: estimators for these models that are robust against high-leverage

7: outliers, which are generally difficult to identify and may cause

8: serious damage to least squares and Huber-type $M$-estimators. For

9: that reason,  robust estimators for semi-functional linear regression

10: models are constructed combining $B$-splines to approximate both the

11: functional regression parameter and the nonparametric component with

12: robust regression estimators based on a bounded loss function

13: and a preliminary residual scale estimator. 

14: Consistency and rates of convergence for the proposed estimators are derived

15: under mild regularity conditions. The reported numerical

16: experiments show the advantage of the proposed methodology

17: over the classical least squares and Huber-type $M$-estimators for

18: finite samples. The analysis of real examples  

19: illustrate that the

20: robust estimators provide better predictions for non-outlying points

21: than the classical ones, and that when potential outliers are removed

22: from the training and test sets both methods behave very

23: similarly.

24: 

25: 

26: \end{abstract}

27: