The active control of vortex shedding from a bluff body consisting of three rotating cylinders in triangular arrangement is studied numerically and experimentally. The flow is characterized at low Reynolds number Re ~ 4000 - 6000. Open-loop cases are considered to understand the influence of maximum rotation rate Ω and the non-dimensional forcing frequency F* scaled by the cylinder radius R and the free-stream velocity U∞. Feedback control is considered by acquiring the hot-wire sensor signal in the wake and using it for the cylinders rotation speeds update. Adaptive and robust controllers are also be designed for the closed-loop control. As the parametric space for the controllers parameters is very large, machine learning using genetic programming will be a major technique for the command law optimization.