The traditional ant colony optimisation (ACO) algorithm, when applied to mobile robot path planning, faces several challenges: slow convergence, susceptibility to local optima, and the generation of paths with excessive turning points, all of which reduce the robot’s operational efficiency. To overcome these shortcomings, this paper proposes a targeted set of improvements designed to enhance algorithm performance and increase the practicality and efficiency of path planning. First, we introduce an initial pheromone enhancement mechanism based on the Bresenham algorithm. By augmenting pheromone concentration along the approximate straight-line path from the start to the goal, ants are guided to explore in the optimal direction, thereby significantly accelerating convergence. Second, we integrate a directional continuity factor into the path selection probability: by using vector dot products to strengthen the bias toward consistent directions and by coupling this with a curvature-based pheromone reward that favours straighter segments, we ensure smoother, more direct paths. Finally, we apply a spring-model-based smoothing strategy as a post-processing step to the paths generated by the ant colony, reducing path complexity and the number of turns to guarantee efficient and reliable robot motion. To validate the performance of the improved algorithm, we conduct comparative experiments on a MATLAB platform against other enhanced ACO variants reported in the literature. The results demonstrate that our proposed algorithm significantly outperforms these existing methods across all performance metrics, exhibiting superior path planning capabilities.