A Hybrid Path Planning Framework for Inspection Robots: Integrating Improved A-Star Algorithm with Dynamic Safety Constraints

Abstract

Path planning is a core technology for enabling inspection robots to operate safely and efficiently in complex industrial spaces, including smart manufacturing sites where sensor reliability, wireless communication continuity, and electromagnetic interference may affect autonomous navigation. Although the classical A-star algorithm is widely used in global path planning because of its completeness and optimality guarantees, it often suffers from computational inefficiency, excessive node expansion, and non-smooth trajectories that are unsuitable for real robot motion. To address these limitations, this work proposes an enhanced A-star algorithm for inspection robots. The heuristic function is redesigned with an adaptive weighting mechanism, balancing search speed and path optimality and reducing the number of expanded nodes. A safety-aware strategy is further introduced by adding virtual safety buffers around obstacles, allowing the robot to maintain a minimum safe distance during navigation. Finally, Bézier curve interpolation is used to improve path smoothness and kinematic feasibility by generating continuous and differentiable trajectories. Simulation experiments show that, compared with the standard A-star algorithm, the proposed method reduces searched nodes by nearly 50% and redundant turns by more than 60%. Overall inspection efficiency is improved by up to 90%, confirming the method’s practicality for robot navigation in complex engineering environments.