A novel 2D motion planning method for vehicles considering the impact of lane configurations

Abstract

Lane changes inherently escalate collision risks, while lane lines mitigate undue cross-lane impacts from lateral perturbations or unsuccessful lane change maneuvers. To quantify these dynamics, this paper introduces an Extended Omnidirectional Risk Indicator (EORI). Building on a novel risk equivalence hypothesis and our previous research, the EORI effectively measures the influence of vehicle relative motion states. A Risk-Quantification based longitudinal planning Model using EORI (ERQM) and an EORI-Based Lane Change model (ELC) are proposed. Unlike conventional models that rely on lane markings to first identify the preceding vehicle, ERQM prioritizes the vehicle presenting the highest risk as the focal object for car-following, allowing it to proactively detect and respond to vehicles that show potential for cutting in. Its mapping relationship between longitudinal steady-state speed and risk offers a novel potential approach for future lane width settings. Besides, ELC dynamically changes the risk search range during the vehicle lane change process, and makes lane change decisions based on EORI. As a model-driven and parameter-free model, ELC enables lane change decisions, duration determination, and trajectory generation. Simulation experiments validate ERQM’s capability to prevent collisions induced by cut-in to a certain extent. Moreover, within the segments selected from the NGSIM dataset, the combination of ERQM and ELC completes lane change with a high success rate, producing more comfortable lane change trajectories. The results demonstrate that EORI effectively represents risk under lane line constraints. The ERQM and ELC models, both based on EORI, adapt well to dynamic multi-participant traffic scenarios, providing a novel model-driven approach for Connected and Automated Vehicles in bidimensional traffic environments.