Applications such as programming spot-welding stations require computing paths in high-dimensional spaces. Random sampling approaches, such as the Probabilistic Roadmaps (PRM) have shown great potential in solving path-planning problems in this kinds of environments. In this paper, we review the description of a new probabilistic roadmap planner, called SBL, which stands for Single-query, Bi-directional and Lazy Collision Checking, and we also add some new results that allow a better comparison of SBL against other similar planners. The combination of features offered by SBL reduces the planning time by large factors, making it possible to handle more difficult planning problems, including multi-robot problems in geometrically complex environments. Specifically, we show the results obtained using SBL in environments with as many as 36 degrees of freedom, and environments in which narrow passages are present.