Initially homogeneous vehicular traffic flow can become inhomogeneous even in the absence of obstacles. We demonstrate how this “phantom jam” phenomenon can be described via traffic models: small perturbations grow into nonlinear traveling waves. These turn out to be analogs of detonation waves in reacting gas dynamics, thus creating a link between traffic flow, combustion, and water roll waves. An outlook is given on the fidelity of data-fitted traffic models, and how these models can provide strategies to control autonomous vehicles so that the fuel consumption of the overall traffic flow is reduced.