This paper addresses the challenge of applying fixed-dimension continuous-domain optimization methods such as the ivy algorithm to autonomous vehicle path planning with discrete constraints and continuity requirements. To this end, we propose a Gaussian-guided algorithm based on an improved ivy optimization method. A Gaussian guidance mechanism is introduced as a bridging modeling approach that iteratively optimizes guide point parameters to steer path generation while leveraging the discrete nature of the problem to substantially reduce computational cost. In addition, the original bias of the ivy algorithm toward low-value solutions is corrected to achieve a more balanced global search, a spiral decay strategy is employed to update the growth rate smoothly for better adaptation to discrete grid environments, and a dynamic time warping (DTW)-based trap escape mechanism is designed to detect stagnation and restore population diversity. Simulation experiments across environments of different scales demonstrate that the proposed method achieves high-quality solutions with improved convergence efficiency, stability, and path smoothness, making it well suited for autonomous navigation tasks.