G 3 . 6 The Evolutionary Planner / Navigator in a mobile robot environment

Based on evolutionary computation concepts, the Evolutionary Planner/Navigator (EP/N) represents a new approach to path planning and navigation. The major advantages of the EP/N include being able to achieve both near-optimality of paths and high planning efficiency, being able to accommodate different optimization criteria, being flexible to changes, and being robust to uncertainties. The EP/N unifies off-line planning and on-line planning/navigation processes in the same evolutionary algorithm to deal with unknowns in an environment gracefully and flexibly. It provides high safety for the robot without requiring complete information about the environment. G3.6.1 Project overview The motion planning problem for mobile robots is typically formulated as follows (Yap 1987): given a robot and a description of an environment, plan a path of the robot between two specified locations which is collision free and satisfies certain optimization criteria. Traditionally there are two approaches to the problem: off-line planning, which assumes a perfectly known and stable environment, and on-line planning, which focuses on dealing with uncertainties when the robot traverses the environment. On-line planning is also referred to by many researchers as the navigation problem. (Although some researchers also interpret navigation as a low-level control problem for path following, we do not use such an interpretation here.) A great deal of research has been done in motion planning and navigation (see Yap 1987 and Latombe 1991 for surveys). However, different existing methods encounter one or many of the following difficulties: • high computation expenses • inflexibility in responding to changes in the environment • inflexibility in responding to different optimization goals • inflexibility in responding to uncertainties • inability to combine advantages of global planning and reactive planning. The EP/N system was developed to address these difficulties; the inspiration to use evolutionary techniques was triggered by the following ideas/observations: • Randomized search can be the most effective in dealing with NP-hard problems and in escaping local minima. • Parallel search actions not only provide great speed but also provide ground for interactions among search actions to achieve even greater efficiency in optimization. • Creative application of the evolutionary computation concept rather than dogmatic imposition of a standard algorithm proves to be more effective in solving specific types of real problems. • Intelligent behavior is the result of a collection of simple reactions to a complex world. • A planner can be greatly simplified, much more efficient and flexible, and increase the quality of search, if search is not confined to be within a specific map structure. • It is more meaningful to equip a planner with the flexibility of changing the optimization goals than the ability of finding the absolutely optimum solution for a single, particular goal. c © 1997 IOP Publishing Ltd and Oxford University Press Handbook of Evolutionary Computation release 97/1 G3.6:1 The Evolutionary Planner/Navigator in a mobile robot environment The EP/N embodies the above ideas by following the evolution program approach, that is, combining the concept of evolutionary computation with problem specific chromosome structures and genetic operators (Michalewicz 1994). With such an approach, the EP/N pursues all the advantages as described above. Less obvious, though, is that, with the unique design of chromosome structure and genetic operators, the EP/N does not need a discretized map for search, which is usually required by other planners. Instead, the EP/N ‘searches’ the original and continuous environment by generating paths based on evolutionary computation. The objects in the environment can simply be indicated as a collection of straight-line ‘walls’. This representation accommodates both known objects and partial information of unknown objects obtained from sensing. Thus, there is little difference between off-line planning and on-line navigation for the EP/N. In fact, the EP/N unifies off-line planning and on-line navigation in the same evolutionary algorithm and chromosome structure. The structure of the EP/N is shown in figure G3.6.1, where FEG—the off-line evolutionary algorithm, and NEG—the on-line evolutionary algorithm—are essentially the same evolutionary algorithm as to be described. The only difference between FEG and NEG is in certain values of parameters (see section G3.6.5) one may choose. The different parameter values are to accommodate slightly different objectives of FEG and NEG: FEG emphasizes the optimality of a path while NEG emphasizes the swiftness in generating a feasible path. Note that both FEG and NEG perform global planning, and NEG generates an alternative subpath by global planning based on the updated knowledge of the environment obtained from sensing. Moreover, if no object is initially known in the environment, then FEG will generate a straight-line path with just two nodes: the start and the goal locations. It will solely depend on the NEG to lead the robot towards the goal while avoiding unknown or newly emerged obstacles.