Interaction-Aware Motion Planning for Autonomous Vehicles With Multi-Modal Obstacle Uncertainty Predictions

Zhou, Jian; Olofsson, Björn; Frisk, Erik · 2024 · Crossref

DOI: 10.1109/tiv.2023.3314709

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 safe and proactive motion planning for autonomous vehicles (EVs) in dynamic, multi-vehicle traffic environments characterized by interaction-aware uncertainties. The authors identify that existing methods often fail to adequately account for the inherent unpredictability of surrounding vehicles (SVs), specifically the multi-modal nature of their potential maneuvers (e.g., lane changes vs. lane keeping) and the trajectory uncertainties within those maneuvers. To bridge this gap, the study proposes an Interaction and Safety-Aware Model Predictive Control (ISA-MPC) framework. This approach integrates interaction-aware motion prediction with robust control strategies to ensure the EV can navigate safely while maintaining performance, even when SV behaviors are uncertain. The methodology combines three key components. First, it utilizes an Interaction-Aware Interacting Multiple Model Kalman Filter (IAIMM-KF) to predict the interactive behaviors of SVs. This filter outputs the probabilities and nominal trajectories for multiple candidate maneuvers for each SV, accounting for vehicle-to-vehicle interactions through a dynamic priority list. Second, the paper introduces a data-driven, online sampling-based method to quantify the trajectory uncertainty for each predicted maneuver. This quantification is derived from real-world traffic data (the highD dataset) and captures the standard deviation of longitudinal and lateral positions, providing a non-conservative estimate of uncertainty. Third, these predictions and uncertainty metrics are integrated into an MPC formulation. The MPC computes an optimal reference trajectory for the EV by incorporating a tunable safety-awareness parameter. This parameter adjusts the predicted obstacle occupancy in the collision-avoidance constraints, allowing the planner to trade off between robustness and performance without increasing computational complexity. The study validates the proposed ISA-MPC through simulations in challenging highway-driving scenarios and using data from recorded traffic datasets. The method is compared against a Scenario MPC (SCMPC), a deterministic MPC, and human-driven vehicle trajectories. Results demonstrate that the ISA-MPC is capable of real-time implementation and effectively handles multi-modal uncertainties. Specifically, the integration of the tunable safety parameter allows the EV to adapt its behavior based on the level of uncertainty in SV predictions, ensuring safety without being overly conservative. The sampling-based uncertainty quantification proved effective in capturing realistic trajectory properties from real-world data, leading to more accurate safety constraints. The significance of this work lies in its holistic approach to handling uncertainty in autonomous driving. By explicitly modeling both maneuver probabilities and trajectory uncertainties within an interaction-aware framework, the proposed method enhances the safety and proactivity of motion planners. The ability to tune the trade-off between performance and robustness via a single parameter offers practical flexibility for deployment in varying traffic conditions. Furthermore, the use of data-driven uncertainty quantification provides a realistic and computationally efficient alternative to conservative worst-case assumptions, contributing to the development of more resilient autonomous systems capable of operating safely in complex, interactive traffic environments.

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.

• situational awareness