Modelling, parameterizing and validating the motion of a tadpole style cargo tricycle with real world experiments

In the past years, cargo bicycles in different configurations have gained popularity for many use cases. Their configurations differ substantially. Single-track cargo bicycles and their kinematics are linked closely to conventional bicycles. The kinematics of inverted tricycles, so-called tadpole trikes, are very different. In this work, we model the motion for such a tadpole tricycle in order to predict the kinematic potential of such a vehicle. A conventional, planar bicycle model is implemented and compared to a planar model that incorporates a term for the lean (or roll) angle. Therefore, the connection between steering and lean angle is calculated by the help of wheel flop. This is validated by inversing the modelling process and optimizing the geometrical approach function with the help of naturalistic cycling studies. The tricycle used for this study is measured experimentally in order to find the parameters for the model. It is then equipped with measuring devices and we present our instrumented tadpole cargo tricycle. By the help of it, we validate the theoretically derived model for the motion of the tadpole tricycle with real world measuring data for given driving scenarios. The models of the tricycle are impinged with data from our experimental driving maneuvers as we do not want to model the behavior of a human driver. It is shown that our derived kinematic models hold reasonably well against the measurements for short term predictions. For the performed driving scenarios, we compare the experimentally measured trajectory with the simulated ones and quantify the error. It is shown that the planar model that incorporated lean performs minimally better compared to the conventional, planar bicycle model. We discuss model limitations as well as potential inaccuracies caused by the used measuring devices on our instrumented cargo tricycle. With the help of the kinematic models, motion prediction of tadpole cargo tricycles can be undertaken. The range, for which the implemented planar models are considered to be valid, is depicted by the range of forward speeds until the liftoff condition. For motion prediction, a conventional, planar bicycle model is sufficient as the more complicated planar model with lean does not substantially outperform it. For maneuvers at the limits of driving dynamics, more sophisticated dynamic models are needed, as the simple kinematic models of this work are not sufficient for this kind of tasks.