The traditional RRT* algorithm suffers from slow convergence, high computational complexity, and unstable path quality. To overcome these problems, this paper proposes an improved AAP-RRT* algorithm (Adaptive Strategies-Artificial Potential Field Integration-Parent Node Construction RRT*). Based the RRT* framework, our approach incorporates four key innovations: (1) adaptive goal-biased sampling and dynamic step size adjustment to reduce random exploration and enhance environmental adaptability; (2) integration of artificial potential field with adaptive attraction and repulsion coefficients to improve goal-orientation and obstacle avoidance; (3) proposing parent node construction to reduce computational overhead while improving path quality; (4) employing greedy node pruning and B-spline smoothing for path optimization. The experimental results demonstrate significant improvements. In complex 3D environment, AAP-RRT* achieves 18.11% shorter paths, 74.39% faster computation, and increases success rates from 62% to 96%. In the real-world testing, our algorithm shows 47.36% shorter paths and 51.05% faster runtime, confirming its practical effectiveness for robotic path planning applications.