The Intelligent Driver Model (IDM), while effective in simulating car-following dynamics for autonomous vehicles (AVs), often produces excessive braking forces during traffic light stops, compromising passenger comfort and energy efficiency. This paper introduces the Blended Acceleration Model (BAM), a novel framework that integrates IDM’s acceleration dynamics with a virtual deceleration function regulated by a dynamic blend factor. BAM adaptively adjusts the deceleration phase based on real-time distance to the traffic light, ensuring smooth transitions between acceleration and braking to mitigate abrupt maneuvers. Tested in a single-traffic-light scenario with a 100 m detection range, BAM combines IDM’s responsiveness with a gradient-based deceleration strategy inspired by stepwise velocity control in automated guided vehicles. The blend threshold factor (α = 10) optimally balances comfort and performance, maintaining deceleration within the comfortable range (≤ 1.5 m/s²) at typical urban speeds (≤ 60 km/h). Simulation results show that BAM achieves a 24.6% reduction in peak deceleration and a 30.6% shorter stopping time compared to IDM, while maintaining well-damped, monotonic deceleration without oscillations. Jerk profiles remain smooth and stable, with temporary peaks occurring only during final braking phases and within acceptable comfort limits. Compared to IDM and the Optimal Velocity Model (OVM), BAM delivers superior braking comfort, smoothness, and stopping accuracy, while OVM, though quicker, induces harsher braking. By addressing IDM’s rigidity in deterministic stopping scenarios, BAM enhances both passenger comfort and operational efficiency, offering strong potential for integration into urban AV control systems.
Loading....