This talk will give insight into the dynamics of turbulent wakes generated by canonical three-dimensional bluff-bodies and their control. Using wind-tunnel measurements (synced PIV and pressure) at Reynolds number of ~200,000 – well into the turbulent regime – we will demonstrate that turbulent coherent structures manifest due to spatiotemporal symmetry break occurring at transitional regimes. Due to ergodicity, these structures restore statistically the broken symmetries at sufficiently high Reynolds numbers, which prevents their identification at turbulent regimes using existing modal decomposition techniques. A symmetry-aware Proper Orthogonal Decomposition framework is developed and implemented based on autoencoders to extract the coherent symmetry-invariant structures. Finally, recent results for the control of coherent structures for road-vehicle drag-reduction applications will be demonstrated. Nonlinear dynamic feedback controllers are optimally tuned using a data-driven Reinforcement Learning framework, capable of stabilising the wake and substantially reducing the mean drag through synthetic jets or flaps placed at the rear trailing edges of the vehicle.