Generic Trajectory Planning Algorithm for Urban Autonomous Driving

Duhautbout, Thibaud; Talj, Reine; Cherfaoui, Veronique; Aioun, Francois; Guillemard, Franck · 2021 · Crossref

DOI: 10.1109/icar53236.2021.9659417

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 presents a generic local trajectory planning algorithm designed for urban autonomous driving, addressing the complexity of dense, dynamic environments with narrow roads and mixed traffic. The authors propose a fully algorithmic method that separates path and speed planning to ensure safety, comfort, and real-time performance. The approach is motivated by the limitations of existing methods, such as lattice planners that struggle with moving obstacles or machine learning approaches that lack interpretability and require extensive data. The proposed architecture integrates global planning, high-level decision-making, and local planning. The local planner utilizes a "parallel tentacles" method to generate candidate paths. These paths consist of a transition segment, defined by fifth-degree polynomials to ensure position, heading, and curvature continuity, followed by a segment parallel to the decision path. Paths are generated using lateral offsets to avoid obstacles and longitudinal offsets to vary transition aggressiveness. The core contribution is a discrete speed adaptation process applied to each path. This process integrates speed constraints based on curvature and decision limits, handles static obstacles and visibility limits by defining stop points with safety distances, and manages moving obstacles by predicting their trajectories and iteratively adjusting the ego-vehicle’s speed profile to avoid collisions. A smoothing algorithm ensures longitudinal comfort by respecting acceleration and deceleration bounds. The algorithm evaluates candidate trajectories by checking for lateral acceleration limits and selecting the optimal path based on safety and comfort metrics. The method is generic, relying on geometric representations rather than specific road shapes, and accounts for the full vehicle shape in collision checks. Simulation results demonstrate that the algorithm produces safe, comfortable, and reactive trajectories in various urban scenarios. The system successfully handles static and dynamic obstacles, including overtaking maneuvers and occluded turns, with processing times compatible with real-time control requirements. The significance of this work lies in its ability to provide a robust, interpretable, and efficient solution for local planning in complex urban settings. By decoupling path generation from speed adaptation and using a geometric approach for obstacle avoidance, the method ensures dynamical feasibility and passenger comfort. The results indicate that the algorithm can effectively anticipate scene evolution and react to fast-changing environments, offering a viable alternative to purely learning-based or static planning methods for autonomous vehicles.

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.

• anticipation