Non-model-based control of a wheeled vehicle pulling two trailers is proposed. It is a fun train for disabled children consisting of a locomotive and two carriages. The fun train has afforded opportunities for both disabled and able bodied young people to share an activity and has provided early driving experiences for disabled children; it has introduced them to assistive and powered mobility. The train is a nonlinear system and subject to nonholonomic kinematic constraints, so that position and state depend on the path taken to get there. The train is described, and then, a robust control algorithm using proportional-derivative filtered errors is proposed to control the locomotive. The controller was not dependent on an accurate model of the train, because the mass of the vehicle and two carriages changed depending on the number, size, and shape of children and wheelchair seats on the train. The controller was robust and stable in uncertainty. Results are presented to show the effectiveness of the approach, and the suggested control algorithm is shown to be acceptable without knowing the exact plant dynamics.