Efficient and safe path planning for a Mobile Robot using genetic algorithm

ROBIL: Robot Path Planning Based on PBIL Algorithm

“…Naderan-Tahan and Manzuri-Shalmani [16] replaced a mutation operator with operators that change (i.e., replace a randomly selected node with an adjacent node in a pre-defined radius), smooth (i.e., eliminate sharp turns) and short-cut (i.e., remove redundant nodes) operators. Yao and Ma [18] and Tuncer and Yildirim [20] also designed new mutation operators by changing the strategy for selecting random nodes for mutation.…”

“…Naderan-Tahan and Manzuri-Shalmani [16] designed a path using a connected Cartesian floating point representation. Shi and Cui [9] modelled a path as a combination of support points derived orthogonally from equidistant segments of the path for obstacle avoidance.…”

“…GAs can also be used to modify the initial path set creation process via feasible population-based GAs (FGAs) [15][16][17]22]. Classical GAs create an initial path set based on randomly selected nodes on the map, whereas FGAs use additional techniques to reconfigure the infeasible paths in the randomly selected initial path set into feasible paths, allowing faster convergence toward a solution set.…”

“…Previous research into applying GAs to path planning problems can be divided into (1) map and path representation [8-14, 16, 17, 19, 20], (2) the design of genetic and problem-specific operators [11][12][13][14][15][16][17][18][19][20], and (3) the creation of an initial feasible path set [15][16][17]22].…”

“…Unfortunately, the performance of these methods depends greatly the upon population size and the number of iterations, as the GA may be unable to find a qualified path within the designated number of iterations or else may end up with a path of relatively low quality. Although researchers have tried to improve conventional GAs by introducing problem-specific operators [11][12][13][14][15][16][17][18][19][20] and initialization techniques [15][16][17][21][22][23], the high number of parameters required by these operators has limited any performance improvements, as well as creating new problems for population initialization.…”