A Potential Field-Based Model Predictive Path-Planning Controller for Autonomous Road Vehicles

Rasekhipour, Yadollah; Khajepour, Amir; Chen, Shih-Ken; Litkouhi, Bakhtiar · 2016 · OpenAlex-citations

DOI: 10.1109/tits.2016.2604240

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 path planning for autonomous road vehicles by integrating artificial potential fields (APF) with model predictive control (MPC). Existing APF methods often neglect vehicle dynamics, while optimal control methods typically treat obstacles and road boundaries as uniform constraints, lacking the flexibility to distinguish between different obstacle types. The authors propose a potential field-based MPC controller that incorporates distinct potential functions for various obstacles and road structures directly into the objective function, allowing the system to prioritize different hazards while ensuring dynamic feasibility. The methodology utilizes a linear bicycle model for vehicle dynamics, discretized for use in the MPC framework. The potential field is constructed using repulsive functions tailored to specific scenarios: hyperbolic functions for non-crossable obstacles (e.g., other vehicles), exponential functions for crossable obstacles (e.g., road bumps), and quadratic functions for lane markers. These potential functions are based on the signed distance between the vehicle and obstacles, normalized by safe longitudinal and lateral distances calculated from vehicle velocity and acceleration limits. The MPC optimization problem minimizes a cost function comprising the potential field value, tracking errors for desired lane and speed, control inputs, and slack variables for soft constraints on tire forces and speed limits. The controller was simulated using a high-fidelity CarSim vehicle model across complicated test scenarios. Results demonstrate that the proposed controller successfully avoids obstacles and adheres to road regulations while maintaining stable vehicle dynamics. Specifically, the system distinguishes between obstacle types; for instance, it navigates around non-crossable obstacles but crosses crossable obstacles like bumps when necessary, prioritizing safety and comfort. The use of a quadratic MPC formulation allows for efficient computation compared to nonlinear alternatives, enabling real-time performance. The significance of this work lies in its ability to combine the flexibility of APF in handling diverse environmental features with the dynamic feasibility guarantees of MPC. By treating obstacles and road structures with distinct potential functions within the optimization objective, the controller can make nuanced decisions based on the nature of the hazard, rather than applying uniform avoidance strategies. This approach improves the practicality of autonomous driving systems by ensuring that planned paths are not only collision-free but also dynamically executable and context-aware.

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