In this paper, we proposed a new learning strategy for probabilistic roadmap (PRM) algorithm. The proposed strategy is based on reducing the dispersion of the generated set of samples. We defined a forbidden range around each selected sample and ignore this region in further sampling. The resulted planner called LD-PRM is an effective multi-query sampling-based planner which is able to solve motion planning queries with smaller graphs. Simulation results indicated that the proposed planner improve the runtime of the PRM algorithm. Furthermore, the proposed planner is able to solve difficult motion planning cases including narrow passages and bug traps, which is a difficult task for classic sampling-based algorithms. For measuring the uniformity of the generated samples, a new algorithm was created to measure the dispersion of a set of samples based on any desired resolution. Also, comparison studies are provided to support the superiority claim of the proposed algorithm.