Planning the Waypoint-Following Task for a Unicycle-Like Robot in Cluttered Environments

Abstract

Abstract:The Introduc onMotion planning for systems with nonholonomic constraints is a topic recently explored by many researchers all around the world. Signi icance of this topic stems from the fact that many wheeled mobile robots and the vast majority of road vehicles are subject to nonholonomic constraints. Ef icient and theoretically sound motion planning in the cluttered environment is paramount to effective automation of wheeled vehicles. While many of the existing motion planners achieve this goal to some extent, they all come with some limitations, which are outlined in the short survey of motion planners presented below. A complete survey of motion planning algorithms can be found in [4,15,17].Availability of powerful computing units has stimulated the development of sampling-based motion planners. Among them, the most successful are Rapidly Exploring Random Trees [13,16] and their variations referred to generally as Rapidly Exploring Dense Trees (see [17]). Those algorithms are applicable to a wide range of systems and perform well in high-dimensional con iguration spaces with no eficient collision checking procedures available. This versatility is achieved by means of probabilistic sampling in the problem space and particular method of extending the tree of possible solutions. Unfortunately, those algorithms are non-deterministic, i.e. two runs with equal boundary conditions may not result in the same solution. Moreover, they are not strictly complete, but rather asymptotically complete, which means that there is no guarantee of inding a solution in inite time. Their Voronoi bias property leads to convergence problems in environments containing narrow passages and in the neighborhood of boundary conditions. Many of those problems can be ixed systematically (see [10]) or heuristically (see [1]) by making speci ic trade-offs. Most of these algorithms rely on numerical simulation of system's evolution for exploring the problem space and make use of no other system-speci ic information.The other major class of motion planners is formed by grid-based algorithms, which model the problem space as a multidimensional lattice. That lattice is searched systematically for a feasible path with algorithms such as Dijkstra, Value Iteration, A* or their variations [2,5]. Effectiveness of the planner depends on choice of the search space and specialization of the search algorithm. Many search spaces were proposed, some of them are: control spaces, con iguration space grids with dynamic neighborhoods [8], state lattices constructed with motion primitives [25] and simplicial decompositions searched continuously [27]. Specialization of search algorithms is achieved mainly by rede inition of movement costs in the grid and simple customization of heuristic functions (see [14]). Some variations of grid-based planners include planners searching in the preprocessed decompositions of the problem space such as hierarchical grids, visibility graphs, etc. Grid-based planners are usually resolution complete, i.e. they will always...

Multi-Strategy Navigation for a Mobile Data Acquisition Platform Using Genetic Algorithms

“…Waypoints are defined as abstract points [15] used to determine local positions [28] through which a mobile platform can navigate, reach its region-of-interest destination, and collect the water pollutant samples [22]. In the application discussed in this paper, waypoints corre-spond to sampling points.…”