Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Physics-Augmented Data-EnablEd Predictive Control for Eco-driving of Mixed Traffic Considering Diverse Human Behaviors

Dongjun Li, Kaixiang Zhang, Haoxuan Dong, Qun Wang, Zhaojian Li, Ziyou Song

TL;DR

The paper tackles eco-driving in mixed traffic by addressing uncertainties from diverse human driver behaviors. It proposes Physics-Augmented DeePC (PA-DeePC), which merges Willems' non-parametric data-driven predictions with partial system physics to improve trajectory forecasting and energy optimization without full parametric models. The approach introduces a physics residual, Hankel-update techniques for handling HDV diversity, and an energy-focused objective that bypasses rigid reference trajectories, yielding measurable energy savings while preserving safety. Simulation results in a five-vehicle platoon demonstrate energy reductions on the order of several percent across a wide set of HDV behaviors, with robust prediction accuracy and safe operation under time-to-collision and time-gap criteria. This work offers a practical, data-efficient framework for improving energy efficiency in mixed CAV/HDV traffic using physics-informed predictive control.

Abstract

Data-driven cooperative control of connected and automated vehicles (CAVs) has gained extensive research interest as it can utilize collected data to generate control actions without relying on parametric system models that are generally challenging to obtain. Existing methods mainly focused on improving traffic safety and stability, while less emphasis has been placed on energy efficiency in the presence of uncertainties and diversities of human-driven vehicles (HDVs). In this paper, we employ a data-enabled predictive control (DeePC) scheme to address the eco-driving of mixed traffic flows with diverse behaviors of human drivers. Specifically, by incorporating the physical relationship of the studied system and the Hankel matrix update from the generalized behavior representation to a particular one, we develop a new Physics-Augmented Data-EnablEd Predictive Control (PA-DeePC) approach to handle human driver diversities. In particular, a power consumption term is added to the DeePC cost function to reduce the holistic energy consumption of both CAVs and HDVs. Simulation results demonstrate the effectiveness of our approach in accurately capturing random human driver behaviors and addressing the complex dynamics of mixed traffic flows, while ensuring driving safety and traffic efficiency. Furthermore, the proposed optimization framework achieves substantial reductions in energy consumption, i.e., average reductions of 4.83% and 9.16% when compared to the benchmark algorithms.

Physics-Augmented Data-EnablEd Predictive Control for Eco-driving of Mixed Traffic Considering Diverse Human Behaviors

TL;DR

The paper tackles eco-driving in mixed traffic by addressing uncertainties from diverse human driver behaviors. It proposes Physics-Augmented DeePC (PA-DeePC), which merges Willems' non-parametric data-driven predictions with partial system physics to improve trajectory forecasting and energy optimization without full parametric models. The approach introduces a physics residual, Hankel-update techniques for handling HDV diversity, and an energy-focused objective that bypasses rigid reference trajectories, yielding measurable energy savings while preserving safety. Simulation results in a five-vehicle platoon demonstrate energy reductions on the order of several percent across a wide set of HDV behaviors, with robust prediction accuracy and safe operation under time-to-collision and time-gap criteria. This work offers a practical, data-efficient framework for improving energy efficiency in mixed CAV/HDV traffic using physics-informed predictive control.

Abstract

Data-driven cooperative control of connected and automated vehicles (CAVs) has gained extensive research interest as it can utilize collected data to generate control actions without relying on parametric system models that are generally challenging to obtain. Existing methods mainly focused on improving traffic safety and stability, while less emphasis has been placed on energy efficiency in the presence of uncertainties and diversities of human-driven vehicles (HDVs). In this paper, we employ a data-enabled predictive control (DeePC) scheme to address the eco-driving of mixed traffic flows with diverse behaviors of human drivers. Specifically, by incorporating the physical relationship of the studied system and the Hankel matrix update from the generalized behavior representation to a particular one, we develop a new Physics-Augmented Data-EnablEd Predictive Control (PA-DeePC) approach to handle human driver diversities. In particular, a power consumption term is added to the DeePC cost function to reduce the holistic energy consumption of both CAVs and HDVs. Simulation results demonstrate the effectiveness of our approach in accurately capturing random human driver behaviors and addressing the complex dynamics of mixed traffic flows, while ensuring driving safety and traffic efficiency. Furthermore, the proposed optimization framework achieves substantial reductions in energy consumption, i.e., average reductions of 4.83% and 9.16% when compared to the benchmark algorithms.
Paper Structure (17 sections, 1 theorem, 19 equations, 6 figures, 2 tables, 1 algorithm)

This paper contains 17 sections, 1 theorem, 19 equations, 6 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Non-Parametric Representation of System Behavior
  4. Data-EnablEd Predictive Control
  5. Physics-Augmented DeePC
  6. System modeling
  7. Linearized Mixed Traffic Model
  8. Energy Consumption Model
  9. Characterization of Human Driver Behaviors
  10. PA-DeePC for eco-driving of mixed traffic
  11. Non-parametric Representation of mixed traffic system
  12. Techniques for Handling Diverse HDVs
  13. Formulation of PA-DeePC for Eco-driving
  14. Simulation results
  15. Experimental Setup
  16. ...and 2 more sections

Key Result

Proposition 1

Willems' fundamental lemma states that if the pre-collected input sequence $u^d_{[1,T]}$ is persistently exciting of order $T_{ini}+N+n$ with $n$ being the dimension of the system states ($n$ can be chosen as an upper bound of state dimension), then at each time step, the patched trajectory $col(u_{ If $T_{ini} \geq$ lag of system, $y$ is uniquely determined from (eq_g), $\forall (u_{ini},y_{ini},

Figures (6)

  • Figure 1: Schematic of the mixed traffic flows. The preceding vehicle is indexed as $0$. The subsequent $n$ vehicles comprise $m$ CAVs and $n-m$ HDVs with indeterminate driving dynamics.
  • Figure 2: Absolute and relative error of the estimated power under different acceleration (i.e., time-varying speed) while $\bar{v}=12 m/s$.
  • Figure 3: Distribution of desired time headway $T_{gap}$.
  • Figure 4: Prediction error during of update and online implementation. The errors are computed via $e_{v} = \widetilde{y}_{v} - \widetilde{v}$, and $e_{s} = \widetilde{y}_{s} - \widetilde{s}$, respectively.
  • Figure 5: Comparison of the energy consumption distribution between two different optimization frameworks under the influence of measurement noise.
  • ...and 1 more figures

Theorems & Definitions (3)

  • Definition 1
  • Proposition 1
  • Definition 2