Human-in-the-loop reinforcement learning

TLDR

Currently, human‑vehicle collaborative driving technologies face significant challenges and lack consideration of a human‑in‑the‑loop learning framework and driving decision‑maker optimization under complex road conditions. The study aims to develop a human‑in‑the‑loop reinforcement learning framework to guide driving decision making and address key challenges in human‑vehicle collaboration. It proposes a hybrid reinforcement learning model and algorithms for both human drivers and autopilots, supported by a driving decision‑maker verification platform. The framework offers a guideline for future research in human‑in‑the‑loop reinforcement learning.

Abstract

This paper focuses on presenting a human-in-the-loop reinforcement learning theory framework and foreseeing its application to driving decision making. Currently, the technologies in human-vehicle collaborative driving face great challenges, and do not consider the Human-in-the-loop learning framework and Driving Decision-Maker optimization under the complex road conditions. The main content of this paper aimed at presenting a study framework as follows: (1) the basic theory and model of the hybrid reinforcement learning; (2) hybrid reinforcement learning algorithm for human drivers; (3)hybrid reinforcement learning algorithm for autopilot; (4) Driving decision-maker verification platform. This paper aims at setting up the human-machine hybrid reinforcement learning theory framework and foreseeing its solutions to two kinds of typical difficulties about human-machine collaborative Driving Decision-Maker, which provides the basic theory and key technologies for the future of intelligent driving. The paper serves as a potential guideline for the study of human-in-the-loop reinforcement learning.