Page 96 - Read Online

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

strategy for integrated global optimization in robot path planning [26] . Lyridis presented an enhanced fuzzy
logic ACO method, demonstrating superior performance to traditional ACO [27] . Hou et al. introduced an
enhanced ACO approach with a communication mechanism, accelerated convergence through an extended
roulette wheel, and designed an adaptive sigmoid decay function to optimize heuristic information in different
stages [28] . Although the research above has improved ACO and achieved preliminary results, they have not
fully considered the maneuverability constraints of UAVs in real-world scenarios. To improve the algorithm’s
ability to search globally, speed up the convergence rate, and generate safe and smooth paths, which will lead
to more efficient UAV path planning that meets practical requirements, it is necessary to optimize the existing
research further.

In this paper, we propose a novel path planning method based on a dual-strategy ACO (DSACO) algorithm.
Our approach centers on optimizing the state transition function and pheromone update rules to enhance the
algorithm’s performance. Firstly, we refine the heuristic factor of the state transition function by incorporating
3Dcharacteristics, whichincludeaddingheuristicinformationregardingthedistancesbetweenadjacentnodes
and the heights of the mountains. Then, a path evaluation function is proposed based on distance, height,
and turning cost. The dynamically adjusted pheromone update strategy helps ants to conduct a global search
in the early stage of the algorithm, accelerates convergence in the later stage of the algorithm, and guides
ants towards the path of the global optimal solution. Doing so effectively steers the ants towards the path
leading to the global optimal solution. Based on the above improvements, a new variation of ACO called the
DSACO algorithm is formed. Subsequently, it is compared with other algorithms based on different terrain
environments. ExperimentalresultsprovethatDSACOhassuperiorglobalsearchcapabilitiesandconvergence
characteristics from four aspects: path length, fitness values, iteration number, and running time.

2. PROBLEM STATEMENT
This paper primarily addresses the issue of static path planning. In this context, static path planning entails
the establishment of an environment model for UAV path planning while simultaneously considering the per-
formance constraints and a comprehensive assessment of the costs associated with the UAV. The ultimate
objective is to pre-plan the path before the UAV embarks on its flight mission.

2.1 3D path planning environment modeling
In static path planning, the UAV’s flight environment can be ascertained before takeoff. Consequently, en-
vironment modeling is vital as it serves as the cornerstone upon which the UAV can base its search for the
optimal path, ultimately facilitating the efficient execution of tasks.

2.1.1 Mountain modeling
This paper studies the problem of UAV path planning in the 3D mountain environment. Given that mountains
can be approximated as cones, the mountainous terrain is characterized by multiple cones with distinct posi-
tions and shapes. We employ a 3D figure described by a natural exponential function with the base number
“e” to elucidate this concept. In this representation, the xOy plane serves as the horizontal reference, and a
point on the mountain is denoted as (x, y, z). The terrain of the natural mountain is described through an
exponential function, as illustrated in Equation (1):

2 2
∑ ( − ) ( − )
− 2 − 2
( , ) = ℎ (1)
=1
Among them, ( , ) represents the height value at the point ( , ), represents the number of peaks in the
mountain environment, ( , ) represents the center coordinates of the peak, ℎ represents the maximum
ℎ
height of the mountain, and ( , ) represents the slope of the mountain. The advantage of simulating
ℎ
mountain peaks with a two-dimensional normal distribution function is that it allows for convenient simula-