Infrastructure Assisted Constrained Connected Automated Vehicle Trajectory Optimization on Curved Roads: A Spatial Formulation on a Curvilinear Coordinate

Ran Yi, Yang Zhou, Xin Wang, Zhiyuan Liu, Xiaotian Li, Bin Ran

TL;DR

This paper introduces a novel spatial formulation for optimizing Connected Automated Vehicle (CAV) trajectories on curved roads, leveraging infrastructure-provided data and a curvilinear coordinate system. This approach allows for more accurate modeling of complex road geometries and dynamic constraints, leading to safer and more efficient driving through a multi-objective Model Predictive Control (MPC) framework.

What Problem Does This Paper Address?

Ensuring that autonomous vehicles can navigate complex road environments safely and efficiently is a critical challenge. Traditional trajectory planning methods often struggle with accurately representing and reacting to spatially varying road features like curvature, lane boundaries, traffic regulations (e.g., speed limits), and dynamic obstacles. Many existing approaches rely on temporal formulations that assume static road conditions, which is unrealistic for curved roads. Furthermore, accurately modeling vehicle kinematics and ensuring smooth, comfortable trajectories while adhering to constraints is difficult. This paper aims to bridge this gap by proposing a method that explicitly incorporates spatial road information and vehicle dynamics into a unified optimization framework.

Methods

The core of this paper's contribution lies in its spatial formulation of trajectory optimization on a curvilinear coordinate system. Instead of relying on time-based planning, the problem is framed in terms of distance traveled along a road's centerline. This allows for a more natural representation of road geometry.

Key methodological aspects include:

  • Curvilinear Coordinate System: The road's centerline is used as a reference, with a coordinate system defined along this path. This simplifies the representation of road boundaries and geometric features.
  • Spatial Formulation: Vehicle states (lateral deviation, angular deviation, pace deviation) and control inputs (steering, acceleration/deceleration) are defined and optimized with respect to distance along the road, rather than time.
  • Infrastructure-Assisted Data: The system assumes that CAVs receive precise, real-time information about road geometry, traffic regulations, and obstacles from roadside units (RSUs).
  • Two-Dimensional Vehicle Kinematics: The model incorporates the vehicle's lateral and longitudinal dynamics within this spatial framework.
  • Multi-Objective Model Predictive Control (MPC): A rolling-horizon MPC is employed to optimize trajectories. This approach allows for systematic handling of multiple objectives (e.g., smoothness, efficiency, safety) and constraints (collision avoidance, vehicle kinematics) in a computationally tractable manner. The MPC is formulated in the discrete spatial domain.
  • Controllability Analysis: The paper rigorously proves the controllability of the linearized system using both linear system theory and small-angle approximations, ensuring that the proposed state-space model can be effectively controlled.

Datasets

Not discussed in the paper. The paper relies on simulated scenarios for evaluation.

Key Insights and Ideas

  • Spatial vs. Temporal Planning: Framing trajectory optimization in a spatial domain, particularly on a curvilinear coordinate, offers significant advantages for handling spatially varying road features compared to traditional temporal approaches. It allows for a more direct and accurate incorporation of road geometry.
  • Infrastructure as a Game-Changer: The reliance on infrastructure-provided data is crucial. It enables the CAV to "see" far ahead and plan proactively, overcoming the limitations of onboard sensors and myopic temporal planning.
  • Curvilinear Coordinates Simplify Geometry: Using the lane centerline as a primary reference greatly simplifies the mathematical formulation of road boundaries and constraints, reducing computational complexity.
  • MPC's Robustness in Spatial Domain: Applying MPC in this spatial formulation proves effective for managing complex constraints and multiple objectives, demonstrating robustness to disturbances like changing road curvature and speed limits.
  • "Buffer Zone" for Smooth Transitions: The introduction of a "buffer zone" for speed limit changes is a practical insight, allowing for smooth, continuous transitions in speed rather than abrupt changes, which is crucial for passenger comfort and safety.

Key Results

The paper presents simulation results across four scenarios:

  1. Continuous Curvy Road (Constant Speed Limit): The proposed method successfully generates smooth trajectories, keeping lateral and angular deviations close to zero after an initial short adjustment period.
  2. Curved Road with Different Driving Behaviors: Demonstrates how tuning the weights in the MPC objective function can lead to either more "obedient" (tightly following the centerline) or "flexible" (smoother, less aggressive) driving behaviors.
  3. Curved Road with an Obstacle: The CAV effectively detects and avoids an obstacle using a flexible driving behavior before and during the obstacle, then reverts to obedient behavior afterward, showcasing safety and adaptability.
  4. Continuous Curvy Road with Speed Limit Change: The "buffer zone" approach successfully manages speed limit changes, resulting in smooth deceleration and speed transitions without compromising trajectory stability.

Across all scenarios, the simulations show that the proposed method provides stable, smooth, and safe trajectories that adhere to constraints and adapt to varying road conditions and disturbances.

Limitations and Future Work

  • Limitations:
    • Point Mass Assumption: The vehicle is treated as a point mass, neglecting its physical dimensions and complex dynamics (e.g., suspension, tire slip).
  • Negligible Delays: Actuation and communication delays are assumed to be negligible, which might not hold in real-world scenarios.
  • Idealized Infrastructure Data: The paper assumes perfect and instantaneous information from infrastructure.
  • Future Work:
    • More Complex Vehicle Models: Incorporating a more detailed four-wheel vehicle model and accounting for uncertainties in vehicle dynamics.
  • Cooperative Planning: Extending the framework to consider the movement of other vehicles and plan trajectories cooperatively.
  • Real-World Testing: While simulations are presented, real-world validation is a natural next step.

Summary updated

Abstract

Vehicle trajectory optimization is essential to ensure vehicles travel efficiently and safely. This paper presents an infrastructure assisted constrained connected automated vehicles (CAVs) trajectory optimization method on curved roads. This paper systematically formulates the problem based on a curvilinear coordinate which is flexible to model complex road geometries. Further, to deal with the spatial varying road obstacles, traffic regulations, and geometric characteristics, two-dimensional vehicle kinematics is given in a spatial formulation with exact road information provided by the infrastructure. Consequently, we applied a multi-objective model predictive control (MPC) approach to optimize the trajectories in a rolling horizon while satisfying the collision avoidances and vehicle kinematics constraints. To verify the efficiency of our method, a numerical simulation is conducted. As the results suggest, the proposed method can provide smooth vehicular trajectories, avoid road obstacles, and simultaneously follow traffic regulations, which is robust to road geometries and disturbances.
BibTeX
@misc{yi2021infrastructure,
  author = {Ran Yi and Yang Zhou and Xin Wang and Zhiyuan Liu and Xiaotian Li and Bin Ran},
  title = {{Infrastructure Assisted Constrained Connected Automated Vehicle Trajectory Optimization on Curved Roads: A Spatial Formulation on a Curvilinear Coordinate}},
  year = {2021},
  eprint = {2103.00699},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2103.00699},
}

Loading related content...

Figure