Decentralized model predictive control for smooth coordination of automated vehicles at intersection

Qian, Xiangjun; Gregoire, Jean; de La Fortelle, Arnaud; Moutarde, Fabien · 2015 · OpenAlex-citations

DOI: 10.1109/ecc.2015.7331068

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of coordinating automated vehicles at intersections without traffic lights, aiming to balance safety, efficiency, and passenger comfort. The authors identify limitations in existing approaches: planning-based methods struggle with computational complexity and robustness to uncertainty, while reactive methods often fail to guarantee deadlock avoidance. Previous work by the authors introduced a priority-based framework that ensures collision-free and deadlock-free maneuvers but relied on a "bang-bang" control law that produced non-smooth, jerky vehicle behaviors and high energy consumption. To resolve this, the paper proposes a decentralized Model Predictive Control (MPC) approach integrated into the priority-based framework, enabling smooth coordination while maintaining provable safety properties. The methodology models vehicle dynamics as second-order integrators constrained to fixed paths. The coordination problem is separated into high-level priority assignment and low-level control. The authors assume a fixed, acyclic priority graph is established, defining the order in which vehicles cross the intersection. The core contribution is a decentralized MPC scheme where each vehicle solves a local optimization problem over a finite horizon. The cost function minimizes the deviation from a target speed and penalizes acceleration magnitude to ensure comfort and fuel efficiency. Crucially, the optimization is subject to "priority-preserving" constraints, requiring the vehicle to remain in a "brake-safe" set relative to higher-priority vehicles. This ensures that if a prior vehicle brakes unexpectedly, the ego vehicle can still avoid collision. To handle the need for predicting prior vehicles' states, the authors propose three strategies: an ideal sequential scheme (MPC*), a constant-velocity prediction scheme (MPC0), and a low-cost cooperative scheme (MPC1) that utilizes shared control inputs from previous time steps. Simulations involving three vehicles demonstrate the effectiveness of the proposed MPC approaches compared to the traditional bang-bang control. The results show that MPC strategies produce significantly smoother velocity and acceleration profiles, allowing vehicles to anticipate conflicts and decelerate earlier rather than reacting abruptly. While the MPC approaches introduce slightly larger traversal delays (approximately 0.5–1.5 seconds) compared to the bang-bang method, they achieve substantial improvements in energy efficiency. Specifically, the MPC strategies reduced fuel consumption by approximately 10% overall. Furthermore, the cooperative MPC1 scheme outperformed the non-cooperative MPC0 scheme for vehicles with multiple predecessors, reducing fuel consumption by an additional 4% due to more accurate predictions of prior vehicles' trajectories. The significance of this work lies in providing a practical, decentralized control solution for autonomous intersections that guarantees safety without sacrificing comfort or efficiency. By integrating MPC with the priority-based framework, the authors demonstrate that it is possible to achieve smooth, energy-efficient driving behaviors while maintaining rigorous collision and deadlock avoidance guarantees. The proposed low-cost cooperation strategy (MPC1) offers a viable path for real-world implementation, balancing computational constraints with the benefits of inter-vehicle communication.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• situational awareness