The path planning of Unmanned Aerial Vehicles (UAVs) is a complex and hard task that can be formulated as a Large-Scale Global Optimization (LSGO) problem. A higher partition of the flight environment leads to an increase in route’s accuracy but at the expense of greater planning complexity. In this paper, a new Parallel Cooperative Coevolutionary Grey Wolf Optimizer (PCCGWO) is proposed to solve such a planning problem. The proposed PCCGWO metaheuristic applies cooperative coevolutionary concepts to ensure an efficient partition of the original search space into multiple sub-spaces with reduced dimensions. The decomposition of the decision variables vector into several sub-components is achieved and multi-swarms are created from the initial population. Each sub-swarm is then assigned to optimize a part of the LSGO problem. To form the complete solution, the representatives from each sub-swarm are combined. To reduce the computation time, an efficient parallel master-slave model is introduced in the proposed parameters-free PCCGWO. The master will be responsible for decomposing the original problem and constructing the context vector which contains the complete solution. Each slave is designed to evolve a sub-component and will send the best individual as its representative to the master after each evolutionary cycle. Demonstrative results show the effectiveness and superiority of the proposed PCCGWO-based planning technique in terms of several metrics of performance and nonparametric statistical analyses. These results show that the increase in the number of slaves leads to a more efficient result as well as a further improved computational time.
Path planning with collision avoidance for unmanned aerial vehicles (UAVs) in environments with moving obstacles is a complex process of navigation, often considered a hard optimization problem. Ordinary resolution algorithms may fail to provide flyable and collision-free paths under the time-consumption constraints required by the dynamic 3D environment. In this paper, a new parallel multiobjective multiverse optimizer (PMOMVO) is proposed and successfully applied to deal with the increased computation time of the UAV path planning problem in dynamic 3D environments. Collision constraints with moving obstacles and narrow pass zones were established based on a mathematical characterization of any intersection with lines connecting two consecutive drones’ positions. For the implementation, a multicore central processing unit (CPU) architecture was proposed according to the concept of master–slave processing parallelization. Each subswarm of the entire PMOMVO population was granted to a corresponding slave, and representative solutions were selected and shared with the master core. Slaves sent their local Pareto fronts to the CPU core representing the master that merged the received set of nondominated solutions and built a global Pareto front. Demonstrative results and nonparametric ANOVA statistical analyses were carried out to show the effectiveness and superiority of the proposed PMOMVO algorithm compared to other homologous, multiobjective metaheuristics.
Path planning plays an important role in autonomous flight control design for Unmanned Arial Vehicles (UAVs). In this paper, a path planning method based on the concepts of Reinforcement Learning (RL) is proposed. A Q-learning algorithm with a dynamic reward function is proposed and successfully implemented to ensure flyable paths with efficient collision avoidance. The dynamic reward function is introduced allowing the UAV to use the real-time distance between its current position and the destination as instructional data during training. Such an improved training mechanism leads to minimize the path length, reduce the learning cost, avoid the inefficient exploration of the search space, and ameliorate the convergence of the Q-planning algorithm. Many performance indicators are considered to evaluate the proposed Q-learning-based UAV's planner. Demonstrative results are implemented to illustrate the efficiency and practicability of the proposed artificial intelligence-based path planning approach.
The increased use of Unmanned Aerial Vehicles (UAVs) in a variety of applications, including illegal activities such as drug trafficking and terrorism, has raised concerns about misuse. To protect special areas against illicit drones operations, UAVs' reconnaissance systems offer a promising solution. Having similar objects in the sky, like birds and airplanes, makes it difficult to detect these UAVs. In this work, a vision-based object recognition system, that uses a deep learning network called You Only Look Once version 5 (YOLOv5), is proposed to recognize drones and distinguish them from birds. The trained model is assessed using a set of performance measures such as mean Average Precision (mAP), precision, recall, F1-score and drone's location detection. A comparison with the YOLOv4 tool is carried out to show the effectiveness and superiority of the proposed YOLOv5 network.