Reactive Motion Planning for Mobile Robots

Motion planning refers to the ability of a system to automatically plan its motions. It is considered central to the development of autonomous robots. In the last decade, much research effort was done on the application of probabilistic roadmaps methods (PRM) for different types of problems (Kavraki et al., 1996; Svestka & Overmars, 1997; Bohlin & Kavraki, 2000; Sánchez & Latombe, 2002). There are two main classes of PRM planners: multiple-query and single-query planners. A multiple-query planner pre-computes a roadmap and then uses it to process many queries (Kavraki et al., 1996; Svestka & Overmars, 1997). In general, the query configurations are not known in advance and the roadmap must be distributed over the entire free configuration space (C-space). On the other hand, a single-query planner computes a new roadmap for each query (Bohlin & Kavraki, 2000; Sánchez & Latombe, 2002; Sánchez et al., 2002). Its only goal is to find a collision-free path between two query configurations. Looking for the smallest space to explore before finding a path. Planners that can answer single queries very quickly and with a little preprocessing are of particular interest. Such planners can be used to re-plan paths in applications where the configuration space obstacles can change. This occurs, for instance, when the robot changes tools, grasps an object, or a new obstacle enters in the workspace. These kinds of planners are more suitable in environments with frequent changes. The adaptation of PRM planners to environments with both static and moving obstacles has been limited so far (Jaillet & Siméon, 2004). The planner proposed by Jaillet and Siméon (Jaillet & Siméon, 2004), uses a combination of single and multiple queries techniques. The proposed planner builds a roadmap of valid paths, considering only the static obstacles, when dynamic changes occurs, the planner uses lazy-evaluation mechanisms combined with a single-query technique as local planner to rapidly update the roadmap. A novel real-time motion planning framework was proposed in (Brock & Kavraki, 2001). It is particularly well suited for planning problems and it decomposes the original planning into simpler sub-problems. The paradigm addresses the planning problems in which a minimum clearance to obstacles can be guaranteed along the solution path. A method for generating collision-free paths for robots operating in changing environments was presented in (Leven & Hutchinson, 2000). The method begins by constructing a graph that represents a roadmap in the configuration space, but this graph is not constructed for a specific workspace. Later, the method constructs the graph for an obstacle-free workspace, and encodes the mapping from workspace cells to nodes and arcs in the graph. When the