We present an intelligent driving software system that is capable of adapting to dynamic task conditions and adjusting to the steering preferences of a human passenger. This interactive autonomous system is a realization of a natural human-robot interaction paradigm known as Adaptation from Participation (AfP). AfP exploits intermittent input from the human passenger in order to adapt the vehicle’s autonomous behaviors to match the intended navigation targets and driving preferences. In our system, this is realized by dynamically adjusting the parameter settings of a vision- based algorithm for tracking terrain boundaries. We deployed the resulting interactive autonomous vehicle in diverse task settings, and demonstrated its ability to learn to drive on unseen paths, as well as to adapt to unexpected changes to the environment and to the vehicle’s camera placement.