In this work, we propose a rapid trajectory planning method based on robotic kinematic constraints, specifically applied to ground-based robot navigation in unknown environments. We address this by formulating a Quadratically Constrained Quadratic Program problem, integrating third-order polynomials and multiple motion states to generate trajectories while considering kinematic constraints. The solutions derived from this method are utilized in initial path point generation. Leveraging the geometric properties of convex decomposition enables path point corrections, while employing Bézier curves enhances trajectory smoothness. We introduce a novel replanning framework encompassing local waypoint generation, trajectory collision detection, and replanning strategies. The viability and effectiveness of our method are demonstrated through simulations and real-world experiments.