To address the shortcomings of the Trigonal Gavialid Colony Optimization Algorithm (TGCOA) in population initialization distribution, temperature strategy selection, and parameter control, this paper proposes an Improved Trigonal Gavialid Colony Optimization Algorithm (ITGCOA). First, a Logistic-Tent hybrid mapping initialization method is introduced, and the Lyapunov exponent is used to guide weight selection, resulting in a more uniform distribution of the initial population in the solution space and enhancing the algorithm’s global exploration capability. Second, a population state feedback-based adaptive temperature selection model is constructed, enabling the algorithm to switch between different strategies according to the search process, thereby balancing global exploration and local exploitation. Finally, a nonlinear adaptive factor is introduced to dynamically adjust the wingbeat amplitude, improving the algorithm’s ability to escape local optima. Simulation experiment results show that ITGCOA has significant advantages in both convergence speed and optimization accuracy, and can plan safer and more optimal flight paths for unmanned aerial vehicles in complex three-dimensional environments.