1: \begin{abstract}                % Abstract of not more than 250 words.

2: Driving under varying road conditions is challenging, especially for autonomous vehicles that must adapt in real-time to changes in the environment, e.g., rain, snow, etc.

3: It is difficult to apply offline learning-based methods in these time-varying settings, as the controller should be trained on datasets representing all conditions it might encounter in the future.

4: %In learning-based autonomous driving methods, the data-driven models used for control are often trained on historical data in road conditions different from the actual condition.

5: While online learning may adapt a model from real-time data, its convergence is often too slow for fast varying road conditions.

6: % Other adaptive methods often require sophisticated models and estimators.

7: We study this problem in autonomous racing, where driving at the limits of handling under varying road conditions is required for winning races.

8: We propose a computationally-efficient approach that leverages an ensemble of Gaussian processes (GPs) to generalize and adapt pre-trained GPs to unseen conditions.

9: Each GP is trained on driving data with a different road surface friction.

10: A time-varying convex % or linear

11: combination of these GPs is used within a model predictive control (MPC) framework, where the model weights are adapted online to the current road condition based on real-time data.

12: The predictive variance % or covariance matrix

13: of the ensemble Gaussian process (EGP) model allows the controller to account for prediction uncertainty and enables safe autonomous driving.

14: Extensive simulations of a full scale autonomous car % Ford Escort %F1/10 autonomous race car 

15: demonstrated the effectiveness of our proposed EGP-MPC method for providing good tracking performance %on surfaces 

16: in varying road conditions and the ability to generalize to unknown maps.

17: \end{abstract}

18: