Efficient and rapid kinodynamic planning is crucial for numerous real-world robotics applications. Various methods have been proposed to address this challenge, primarily falling into two categories: (a) randomized planners and (b) trajectory optimization utilizing simplified models and numerical optimization. Randomized planners such as RRT and PRM excel in exploring the state space, while trajectory optimization methods, like direct collocation, are adept at discovering optimal trajectories within well-defined spaces. We aim to achieve effective and efficient kinodynamic planning and exploration by integrating evolutionary algorithms (Quality-Diversity) with trajectory optimization. Our preliminary experiments showcase that using the proposed methodology we get the best from both worlds on two simulated experiments.