[Paper Review] Combining Social Force Model with Model Predictive Control for Vehicle's Longitudinal Speed Regulation in Pedestrian-Dense Scenarios

Dongfang Yang, Ümi̇t Özgüner|arXiv (Cornell University)|Jul 11, 2019

Traffic control and management8 references4 citations

TL;DR

This paper proposes a model predictive control (MPC) strategy combined with a social force-based vehicle-crowd interaction (VCI) model to regulate autonomous vehicle speed in pedestrian-dense scenarios. By predicting pedestrian motion and formulating a quadratic programming (QP) optimization problem with safety and speed objectives, the method achieves safer, faster navigation than traditional PID control, especially in high-density crowds with up to 1.27 seconds faster completion time on average.

ABSTRACT

In pedestrian-dense traffic scenarios, an autonomous vehicle may have to safely drive through a crowd of pedestrians while the vehicle tries to keep the desired speed as much as possible. This requires a model that can predict the motion of crowd pedestrians and a method for the vehicle to predictively adjust its speed. In this study, the model-based predictive control (MPC) was combined with a social-force based vehicle-crowd interaction (VCI) model to regulate the longitudinal speed of the autonomous vehicle. The predictive feature of the VCI model can be precisely utilized by the MPC. A criterion for simultaneously guaranteeing pedestrian safety and keeping the desired speed was designed, and consequently, the MPC was formulated as a standard quadratic programming (QP) problem, which can be easily solved by standard QP toolbox. The proposed approach was compared with the traditional proportional-integral-derivative (PID) control approach for regulating longitudinal speed. Scenarios of different pedestrian density were evaluated in simulation. The results demonstrated the merits of the proposed method to address this type of problem. It also shows the potential of extending the method to address more complex vehicle-pedestrian interaction situations.

Motivation & Objective

• To address the challenge of autonomous vehicle longitudinal speed regulation in unsignalized, pedestrian-dense environments where safety and desired speed must be balanced.
• To improve upon traditional PID control by incorporating predictive capability and pedestrian motion modeling.
• To design a control strategy that ensures pedestrian safety while minimizing delays during vehicle-pedestrian interactions.
• To formulate the MPC problem as a standard quadratic programming (QP) problem for efficient, real-time solvability.

Proposed method

• The vehicle-crowd interaction (VCI) model uses social forces to simulate pedestrian motion, including repulsion from vehicles, with a modified vehicle influence model for better accuracy.
• The MPC framework uses predictions of pedestrian motion over a finite horizon to optimize vehicle speed in real time.
• A custom cost function balances maintaining the desired speed and minimizing control effort, while state constraints enforce a safe distance from the closest pedestrian.
• The MPC problem is converted into a standard quadratic programming (QP) problem, enabling efficient solution via standard QP solvers.
• The VCI model is integrated into MPC at each time step to provide dynamic, predictive pedestrian motion forecasts.
• The control action is updated iteratively based on the QP solution, enabling real-time adaptation to changing pedestrian flows.

Experimental results

Research questions

• RQ1 Can a predictive control strategy that combines MPC with a social force-based VCI model effectively regulate vehicle speed in high-density pedestrian scenarios?
• RQ2 How does the proposed MPC-VCI method compare to traditional PID control in terms of safety, speed maintenance, and interaction time?
• RQ3 To what extent does the predictive capability of the VCI model enhance control performance compared to non-predictive methods?
• RQ4 How does pedestrian density affect the performance gap between MPC and PID control in vehicle-pedestrian interaction?

Key findings

• The MPC approach reduced the average interaction time by 1.27 seconds compared to PID in scenarios with 30 pedestrians, indicating significantly faster and more efficient navigation.
• In stop-and-wait situations, the MPC approach reduced the maximum waiting time by 1.95 seconds on average, demonstrating superior delay management.
• In non-stop scenarios, the MPC approach reduced interaction time by 0.98 seconds on average, showing consistent performance improvement.
• The PID controller exhibited a steady-state error of approximately 0.16 m/s at the target speed of 4 m/s, contributing to a maximum delay of 0.4 seconds in interaction completion.
• Performance degradation was observed in low-density scenarios (10–20 pedestrians), but MPC still outperformed PID in all tested cases.
• The proposed method successfully transformed the MPC problem into a standard QP, enabling reliable and efficient real-time solution using standard QP toolboxes.