Robust exploration in linear quadratic reinforcement learning