Traditional risk map modeling for robot motion control often computes risk values across the entire map and sends all potential collision areas to the planner. This causes two problems: high computational load in high-Degree of Freedom(DOF) systems and limited real-time perception in real environments. To address this, we propose a Riskmap model for complex dynamic environments. Static and dynamic obstacles are modeled using 2D Gaussian distributions. For dynamic obstacles, risk is distributed only in the movement direction using an elliptical Gaussian cross-section. A forward perception mechanism further refines the map locally with an elliptical Gaussian focused on the robot’s position and a forward point. This improves both accuracy and efficiency. Simulations show the model enhances navigation safety and real-time performance in dynamic, cluttered environments.