In order to solve the problems of low optimization accuracy and weak obstacle avoidance ability of agricultural unmanned aerial vehicle (UAV) when carrying out path planning tasks in the environment of obstacle interference, a wireless ultraviolet(UV) light flight path guidance system was established by using the characteristics of all-weather non-direct communication, covert communication and various special occasions of wireless ultraviolet light communication guidance equipment. An improved bat algorithm based on exponential decline strategy (EDS-IBA) was proposed. The dynamic inertia weight based on exponential decline strategy and the artificial potential field of adjacent path points was introduced to ensure the smooth path of the UAV while avoiding falling into the local optimum. The diversity and authenticity of the flight environment were guaranteed through the cost of obstacle interference and the cost of ultraviolet communication. The results show that, compared with the improved bat algorithm based on differential evolution algorithm and the traditional bat algorithm, in 2-D environment, the average path length of EDS-IBA is shortened by 10.7% and 16.3% respectively; in 3-D environment, the planned average path is shortened by 13.7% and 36.2% respectively. When the algorithm reaches the convergence state, the fitness value of EDS-IBA is also small, which has good feasibility and effectiveness in path planning ability and algorithm performance. This research has certain practical significance for the practical application and development of UAV path planning.