A Review of Tracking, Prediction and Decision Making Methods for Autonomous Driving

TL;DR

This review paper comprehensively examines current methods for tracking, predicting, and decision making in autonomous driving, highlighting neural network and reinforcement learning approaches for improving vehicle navigation and safety.

Contribution

It provides a detailed overview of recent advances in tracking, prediction, and decision-making techniques, emphasizing neural network and stochastic methods used in autonomous vehicles.

Findings

Neural network-based tracking and prediction methods are widely used.

Deep reinforcement learning enhances decision-making in autonomous driving.

Monte Carlo Tree Search aids in exploring alternative driving actions.

Abstract

This literature review focuses on three important aspects of an autonomous car system: tracking (assessing the identity of the actors such as cars, pedestrians or obstacles in a sequence of observations), prediction (predicting the future motion of surrounding vehicles in order to navigate through various traffic scenarios) and decision making (analyzing the available actions of the ego car and their consequences to the entire driving context). For tracking and prediction, approaches based on (deep) neural networks and other, especially stochastic techniques, are reported. For decision making, deep reinforcement learning algorithms are presented, together with methods used to explore different alternative actions, such as Monte Carlo Tree Search.