Abstract:Autonomous driving systems face the formidable challenge of navigating intricate and dynamic environments with uncertainty. This study presents a unified prediction and planning framework that concurrently models short-term aleatoric uncertainty (SAU), long-term aleatoric uncertainty (LAU), and epistemic uncertainty (EU) to predict and establish a robust foundation for planning in dynamic contexts. The framework uses Gaussian mixture models and deep ensemble methods, to concurrently capture and assess SAU, LAU, and EU, where traditional methods do not integrate these uncertainties simultaneously. Additionally, uncertainty-aware planning is introduced, considering various uncertainties. The study's contributions include comparisons of uncertainty estimations, risk modeling, and planning methods in comparison to existing approaches. The proposed methods were rigorously evaluated using the CommonRoad benchmark and settings with limited perception. These experiments illuminated the advantages and roles of different uncertainty factors in autonomous driving processes. In addition, comparative assessments of various uncertainty modeling strategies underscore the benefits of modeling multiple types of uncertainties, thus enhancing planning accuracy and reliability. The proposed framework facilitates the development of methods for UAP and surpasses existing uncertainty-aware risk models, particularly when considering diverse traffic scenarios. Project page:

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to handle uncertainty when navigating in complex and dynamic environments in an autonomous driving system. Specifically, the research proposes a unified prediction and planning framework that simultaneously models short - term aleatoric uncertainty (SAU), long - term aleatoric uncertainty (LAU) and epistemic uncertainty (EU) to predict and establish a robust foundation for planning in dynamic environments. Traditional methods usually do not integrate these uncertainties simultaneously, while the framework proposed in this paper uses Gaussian mixture models and deep integration methods to simultaneously capture and evaluate SAU, LAU and EU. The main contributions of the paper include: 1. Defining and modeling three types of uncertainties in prediction: A new integrated network architecture is proposed, which can simultaneously model all or part of SAU, LAU and EU, enhancing the reliability of prediction and providing more abundant information and a more robust foundation for planning in dynamic environments. 2. Uncertainty - aware planning (UAP): A prediction - planning framework considering various uncertainties is proposed. Risk models for different uncertainties are designed, enabling the planning module to consider multiple uncertainty factors, ensuring that the prediction - planning system effectively copes with dynamic and uncertain real - driving scenarios and the potential functional limitations inherent in the prediction module. 3. Comprehensive comparative analysis of various uncertainties and their roles in planning: Based on the proposed prediction - planning framework, comprehensive experiments are designed to compare the advantages of different uncertainties, risk models and UAP methods, providing a clear and systematic perspective and inspiration. Through these contributions, this research aims to fill the gap in the existing research on comprehensive modeling and comparison of different types of uncertainties, and provide valuable insights into how various uncertainties and their combinations affect the reliability of autonomous driving systems.

Uncertainty-Aware Prediction and Application in Planning for Autonomous Driving: Definitions, Methods, and Comparison

Uncertainty-Aware Motion Planning for Autonomous Driving on Highway

Prediction-Uncertainty-Aware Decision-Making for Autonomous Vehicles

Neural Network Based Uncertainty Prediction for Autonomous Vehicle Application.

An Uncertainty-Aware Lane Change Motion Planning Algorithm Based on Probabilistic Trajectory Prediction Distribution

Integrated Decision Making and Planning Based on Feasible Region Construction for Autonomous Vehicles Considering Prediction Uncertainty

Drive with Quantified Risk: A Probabilistic Approach for Occlusion-Aware Trajectory Planning

Integrated Decision Making and Planning Framework for Autonomous Vehicle Considering Uncertain Prediction of Surrounding Vehicles

Safe planning and control under uncertainty for self-driving

Safety-Assured Speculative Planning with Adaptive Prediction

Multi-Uncertainty Aware Autonomous Cooperative Planning

Manoeuvre Prediction and Planning for Automated and Connected Vehicles Based on Interaction and Gaming Awareness under Uncertainty

Feature Uncertainty Estimation In Sensor Fusion Applied To Autonomous Vehicle Location

Uncertainty-Aware Model-Based Reinforcement Learning: Methodology and Application in Autonomous Driving

Uncertainty-Aware Model-Based Reinforcement Learning with Application to Autonomous Driving

Unravelling uncertainty in trajectory prediction using a non-parametric approach

Uncertainty-Aware Decision-Making and Planning for Autonomous Forced Merging

A Possibilistic Risk Assessment Framework for Unmanned Electric Vehicles with Predict of Uncertainty Traffic

Failure Detection for Motion Prediction of Autonomous Driving: An Uncertainty Perspective

Long-Tail Prediction Uncertainty Aware Trajectory Planning for Self-driving Vehicles.

Prediction Failure Risk-Aware Decision-Making for Autonomous Vehicles on Signalized Intersections