This article introduces a model-based reinforcement learning (RL) approach for continuous state and action spaces. While most RL methods try to find closed-form policies, the approach taken here employs numerical on-line optimization of control action sequences. First, a general method for...

Let Ξ0=[−1,1], and define the segments Ξn recursively in the following manner: for every n=0,1,…, let Ξn+1=Ξn∩[an+1−1,an+1+1], where the point an+1 is chosen randomly on the segment Ξn with uniform distribution. For the radius ρn of Ξn, we prove that n(ρn−1/2) converges in...