A NOVEL PATH PLANNING FOR AUV BASED ON DUNG BEETLE OPTIMISATION ALGORITHM WITH DEEP Q-NETWORK

This study presents a route planning technique, known as dung beetle optimisation with deep Q-network (DBO-DQN), to tackle the difficulties associated with quick path planning and efficient obstacle avoidance for autonomous underwater vehicles (AUVs) operating in a 3D underwater environment. A reinforcement learning approach has been devised to enhance the convergence of DBO. The proposed approach involves substituting the uniformly distributed random number in the updating function with a randomly generated number drawn from a specific normal distribution. The estimation of the mean and standard deviation of the normal distribution is achieved by utilising the present state of each person through the DQN algorithm. The utilisation of the piecewise logistic chaotic mapping initialises the population with the objective of enhancing the variety of the population. In conclusion, taking into account the unique characteristics of the underwater environment, a fitness function is formulated that incorporates both the length of the route path and the deflection angle. This enables the algorithm to identify a solution path that minimises energy usage in the underwater environment. The efficacy of the suggested strategy in comparison to conventional methods is proven by both simulation and experimental findings.