Vehicle speed controls, as adaptive cruise control and related automated evolutions, are control systems able to follow a desired vehicle reference speed that is set by the driver and fused with information as road signs, SD maps etc.. Current normal production systems don’t distinguish among the vehicle users, only some carmakers are doing first steps towards the introduction of learning from driver to adapt the traditional control. In our work, we follow up this content with a humanized speed control, based on learning of driver longitudinal behavior. This method is able to combine machine learning algorithms, vehicle positioning and recurrent trips into existing automated longitudinal control systems. Proposed algorithm can reduce the interactions between drivers and automated systems by improving the acceptance of automated longitudinal control. Furthermore, proposed integration works mainly on speed reference that dramatically simplifies the customization of the system. We present the general methodology of our online learning procedure and suggest how to integrate proposed work in a normal production vehicle.
