Page 95 - Read Online

Page 2 of 19 Mai et al. Intell Robot 2023;3(4):466-84 I http://dx.doi.org/10.20517/ir.2023.37

1. INTRODUCTION
Duetotheiruniquetechnicalattributes, unmannedaerialvehicles(UAVs)canbeequippedwithdiversesensor
devices to achieve real-time environmental monitoring, self-position determination, continuous flight attitude
adjustments, and obstacle avoidance. Consequently, UAVs exhibit a remarkable capacity to accomplish their
[1]
designated tasks efficiently . The primary objective of UAV path planning is to find an optimal and feasible
path within a known environment that is free of conflicts and meets the optimization criteria, given predefined
[2]
start and end point locations . In real-world settings, UAV flight missions are subject to many uncontrollable
factors due to the inherent uncertainty and dynamics of the environment. Hence, research on UAV path
planning holds profound practical significance.

Scholars have developed a plethora of algorithms to address the problem of two-dimensional path planning,
[3]
[5]
[4]
suchastheelementdecompositionmethod ,thepotentialfieldmethod ,andDijkstra’salgorithm ,among
others. Nevertheless, these algorithms typically overlook height constraints and may not align with the practi-
cal flight requirements of real UAVs. Therefore, three-dimensional (3D) path planning, which systematically
accountsformanyconstraints,hasemergedasthecentralresearchfocusinpathplanning. Wuetal. introduced
a multi-step A* search algorithm for offline and online path planning for UAVs in a four-dimensional context,
[6]
encompassingthreespatialandtemporaldimensions . Shorakaeietal. devisedapathplanningmethodology
that leverages probability graphs, integrating them with genetic algorithms and introducing novel genetic oper-
[7]
ators to select apt chromosomes for crossover operations . Roberge et al. proposed using genetic algorithms
and particle swarm algorithms to solve autonomous UAV path planning problems in complex 3D environ-
ments, taking into account the width of the UAV and the optimal trajectory criterion in 3D environments to
[8]
reduce the execution time of the solution . Abeywickrama et al. presented an artificial potential field model
[9]
that demonstrates remarkable efficiency in reducing collisions among UAVs . Vanegas et al. proposed a
method for optimizing 3D UAV path planning using a non-holonomic constraint path planning approach [10] .
Lastly, Jain et al. introduced an innovative algorithm based on the Multiverse Optimizer algorithm to en-
hance the time efficiency and precision of UAV path planning within a 3D environment [11] . This approach
incorporates the Munkres algorithm into UAV path planning, further augmenting its effectiveness. Wang et al.
introduced an optimized list-based simulated annealing (LBSA) algorithm tailored to address the challenges
posed by the large-scale traveling salesman problem (TSP) [12] . Li presented a refined tabu search algorithm
incorporating a greedy algorithm for addressing the random vehicle routing problem [13] . Kala et al. integrated
a fuzzy inference system with an A* algorithm to address challenges in robot path planning [14] . Since Dorigo
et al. proposed ant colony optimization (ACO), it has gradually been applied in logistics and path planning [15] .
The algorithm benefits from strong robustness and good information feedback by imitating the principle of
ant colony foraging, which helps solve the challenge of complex path planning. In the 1990s, the prominent
representatives of ACO algorithms were the Ant System (AS) [16] and the two most successful variants: MAX-
MIN Ant System (MMAS) [17] and Ant Colony System [18] . ACO algorithms have constantly been modified
and extensively developed up to this day. Li et al. used the geometric optimization method to guide the ants,
accelerating the convergence speed [19] . However, there was an issue with individual ants becoming disori-
ented. Literature [20,21] combines self-adaptation and ACO algorithms to improve the algorithm’s capability
to find the global optimum through adaptive parallelism and information updating strategies. Despite these
advancements, it is worth noting that the resulting path generated by this algorithm still exhibits non-smooth
characteristics. Chen et al. incorporated Poisson distribution to simulate the influence of unknown factors
and established a three-color raster map [22] . The improved algorithm can design the optimal route safely and
effectively in the environment under the influence of unknown factors. Yi et al. introduced a multi-factor
heuristic function strategy to improve ACO’s global search capability and convergence [23] . Ning et al. de-
signed an enhanced pheromone update mechanism based on ACO, strengthening pheromones on edges and
enhancing global search capabilities and convergence [24] . Wang et al. transformed ACO into a Time-Sensitive
Network (TSN), resulting in better convergence speed, optimization ability, and reduced susceptibility to lo-
cal optima compared to traditional ACO [25] . Miao et al. proposed an improved Adaptive ACO (IAACO)