A Hybrid Planning Approach Based on MPC and Parametric Curves for Overtaking Maneuvers

Lattarulo, Ray; Rastelli, Joshué Pérez · 2021 · OpenAlex-citations

DOI: 10.3390/s21020595

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 executing safe and comfortable overtaking and obstacle avoidance maneuvers in Automated Driving Systems (ADS). While ADS technology has advanced, it often lacks the capacity to handle risky scenarios like lane changes without compromising passenger comfort or safety. The authors propose a "Hybrid Planning" approach that combines the smoothness of parametric Bézier curves for nominal trajectory generation with the reactive capabilities of Model Predictive Control (MPC) for handling unexpected conditions. This method aims to resolve the limitations of existing sampling-based methods, which struggle with complex dynamic environments, and pure optimization methods, which may lack computational efficiency or generality. The methodology employs a decoupled linear model for the MPC formulation to ensure short computation times. The longitudinal dynamics are modeled using a triple integrator chain based on jerk, while lateral dynamics use a double integrator chain. This linear approach allows for fast real-time optimization via Quadratic Programming, with constraints applied to acceleration, jerk, speed, and lateral offset to ensure comfort and safety. The system utilizes Vehicle-to-Everything (V2X) communication to obtain real-time data on obstacles and other vehicles. The hybrid planner merges the pre-computed Bézier nominal trajectory with the MPC-generated lateral offset and longitudinal speed profile. A collision verification module evaluates future vehicle states against propagated obstacle positions, dynamically adjusting lateral bounds to switch between the nominal lane and the opposite lane if a collision is predicted. The approach was validated through real-world tests using an automated Renault Twizy vehicle. The experiments involved complex scenarios with both static and moving obstacles at speeds up to 60 kph. The results demonstrated that the hybrid planning method successfully executed overtaking maneuvers while maintaining trajectory smoothness and avoiding collisions. The system effectively reacted to unexpected conditions by adjusting the lateral offset and speed profile in real-time, leveraging the V2X data to predict conflicts and modify the trajectory accordingly. The tests confirmed that the linear MPC model provided sufficient computational performance for real-time operation without the need for heavy non-linear solvers. The significance of this work lies in its ability to balance passenger comfort with safety in dynamic driving environments. By integrating Bézier curves for smooth nominal paths with MPC for reactive maneuvering, the approach offers a robust solution for tactical and operational level decision-making in autonomous vehicles. The use of a decoupled linear model ensures that the system can handle real-time constraints and unexpected obstacles efficiently. This hybrid strategy provides a generalizable framework for overtaking and obstacle avoidance that does not rely on hard-coded specific cases, thereby enhancing the reliability and safety of automated driving systems in complex traffic conditions.

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.

• automation surprise