Safe Merging of Autonomous Vehicles under Temporal Logic Task

Zhao, Chenguang; Yu, Huan · 2024 · Unknown

DOI: 10.1109/cdc56724.2024.10886277

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 controlling autonomous vehicles (AVs) during merging maneuvers at highway ramps, a scenario prone to congestion and accidents. The authors focus on a specific triplet scenario where an AV merges from a finite-length acceleration lane into mainline traffic between a leading human-driven vehicle (LV) and a following human-driven vehicle (FV). The primary motivation is to overcome the limitations of existing control methods: Model Predictive Control (MPC) suffers from high computational loads and inaccurate future state predictions, while standard Control Barrier Functions (CBF) fail to enforce time-critical merging constraints. To resolve this, the paper proposes a novel controller that integrates safety, task-specific, and physical constraints using Signal Temporal Logic (STL) and time-varying CBFs. The methodology involves modeling the longitudinal dynamics of the three vehicles and defining three distinct constraint categories expressed in STL syntax. First, a task constraint ensures the merge is completed within the acceleration lane’s length and a maximum duration, formulated as an "eventually" STL task. Second, safety constraints require the AV to maintain a safe gap with both the LV and FV upon merging, based on Time-to-Collision (TTC) policies, formulated as "always" STL tasks. Third, physics constraints enforce positive speed and a maximum speed limit. The authors translate these STL formulas into time-varying CBFs, which provide input constraints for a quadratic programming solver. This approach allows the controller to satisfy complex temporal logic requirements without the need for future state prediction, thereby reducing computational complexity compared to MPC. The study validates the proposed controller through simulations using 95 real-world merging trajectories extracted from the Next Generation Simulation (NGSIM) dataset. The results demonstrate that the controller successfully satisfies all defined safety, task, and physical constraints. Specifically, the AV completes the merging maneuver safely within the spatial and temporal limits. Furthermore, the controller yields smoother traffic flow with less speed perturbation compared to baseline methods. The simulations also indicate that the proposed method reduces the time required to complete the merge, thereby enhancing overall traffic efficiency. The significance of this work lies in its introduction of the first merging controller design based on STL, bridging the gap between formal methods and real-time control. By encoding temporal logic into time-varying CBFs, the approach provides rigorous theoretical guarantees for safety and task completion while maintaining low computational loads. This method offers a robust solution for mixed traffic scenarios, ensuring that autonomous vehicles can merge safely and efficiently without relying on uncertain predictions of human-driven vehicle behavior.

Key finding

The proposed signal temporal logic and control barrier function-based controller enables autonomous vehicles to merge safely and efficiently within temporal and spatial constraints, outperforming model predictive control in computational load and robustness.

Methodology

simulation_modeling

Sample size: 95

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. Discovered via author_sweep_intake on 2026-05-28.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

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

• gap acceptance