Capture Uncertainties in Deep Neural Networks for Safe Operation of Autonomous Driving Vehicles

TL;DR

This paper introduces a framework for safe autonomous driving that quantifies and propagates perception and motion uncertainties in deep neural networks, enhancing safety without sacrificing efficiency.

Contribution

It presents a Bayesian DNN for 3D object detection with uncertainty quantification and an uncertainty-aware motion planning algorithm, PU-RRT, for safer autonomous vehicle operation.

Findings

Improved safety in simulated complex scenarios

Effective uncertainty quantification in perception and motion

Enhanced robustness of autonomous driving under uncertainty

Abstract

Uncertainties in Deep Neural Network (DNN)-based perception and vehicle's motion pose challenges to the development of safe autonomous driving vehicles. In this paper, we propose a safe motion planning framework featuring the quantification and propagation of DNN-based perception uncertainties and motion uncertainties. Contributions of this work are twofold: (1) A Bayesian Deep Neural network model which detects 3D objects and quantitatively captures the associated aleatoric and epistemic uncertainties of DNNs; (2) An uncertainty-aware motion planning algorithm (PU-RRT) that accounts for uncertainties in object detection and ego-vehicle's motion. The proposed approaches are validated via simulated complex scenarios built in CARLA. Experimental results show that the proposed motion planning scheme can cope with uncertainties of DNN-based perception and vehicle motion, and improve the operational safety of autonomous vehicles while still achieving desirable efficiency.