Path planning for unmanned aerial vehicles in complex environment based on an improved continuous ant colony optimisation

Abstract

To address the complex challenge of unmanned aerial vehicle (UAV) path planning, a novel continuous ant colony optimisation with an improved state transition probability, a random-walk strategy and an adaptive waypoints-repair method (ACOSRA_R) is proposed to enhance the efficiency and accuracy of UAV 3D path planning. In ACOSRA_R, an improved state transition probability is integrated to simplify the search process, enabling the algorithm to converge rapidly. A random-walk strategy involves switching between employing Brownian motion and Lévy flight to help it escape from local optima in the later stage and increase the possibility of exploring new solutions. An adaptive waypoints-repair method is proposed to repair waypoints in the infeasible domain to enhance flight efficiency. To validate its performance, ACOSRA_R is compared with seven advanced meta-heuristic algorithms on 9 real digital elevation model maps. Experimental results show that ACOSRA_R outperforms other comparison algorithms, efficiently generating higher-quality UAV paths in different environments. Additionally, we successfully integrated the dynamic window approach with ACOSRA_R to solve UAV path planning in a partially unknown scenario with static and moving obstacles.