Reinforcement-Learning-Based Decision and Control for Autonomous Vehicle at Two-Way Single-Lane Unsignalized Intersection

Abstract

Intersections have attracted wide attention owing to their complexity and high rate of traffic accidents. In the process of developing L3-and-above autonomous-driving techniques, it is necessary to solve problems in autonomous driving decisions and control at intersections. In this article, a decision-and-control method based on reinforcement learning and speed prediction is proposed to manage the conjunction of straight and turning vehicles at two-way single-lane unsignalized intersections. The key position of collision avoidance in the process of confluence is determined by establishing a road-geometry model, and on this basis, the expected speed of the straight vehicle that ensures passing safety is calculated. Then, a reinforcement-learning algorithm is employed to solve the decision-control problem of the straight vehicle, and the expected speed is optimized to direct the agent to learn and converge to the planned decision. Simulations were conducted to verify the performance of the proposed method, and the results show that the proposed method can generate proper decisions for the straight vehicle to pass the intersection while guaranteeing preferable safety and traffic efficiency.

Prophetic Energy Assessment with Smart Implements in Hydroelectricity Entities Using Artificial Intelligence Algorithm

“…To examine the effect of applied optimization in the energy management systems of different devices a transportation vehicle is introduced where decisions are taken automatically with best control practices [21].…”

“…Even the above-mentioned case studies are analyzed by several researchers in Morocco where energyefficient operation is achieved only if AI is incorporated. All the existing models [21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36] focuses only on different renewable energy sources to forecast various behavior of appliances in real-time environmental conditions. But most of the procedures that are present in existing methodologies are not introduced with high-end monitoring devices and even the analytical framework is not framed.…”

“…is can be proved with 15 different appliances as the proposed method crosses the robustness line without any marginal limit but in the next case with 16 appliances after reducing the number of renewable sources the projected method provides low robust conditions. But in the existing method [21] if the number of appliances is increasing then robustness of all installed appliances increases in such a way thus energy management is not guaranteed as renewable sources are increased in the system.…”

Intelligent Control of Robots with Minimal Power Consumption in Pick-and-Place Operations

“…Together with unsupervised and supervised learning approaches, deep reinforcement learning (RL) is now considered the most progressive machine-learning methodology, during which the system is forcibly trained using a "stimulus-response" principle [37][38][39][40][41]. Unlike the previous two, RL does not rely on a static dataset, but operates in a dynamic environment and learns from collected experiences.…”

A Method for High-Value Driving Demonstration Data Generation Based on One-Dimensional Deep Convolutional Generative Adversarial Networks

“…The introduction of fixed Q-targets and experience replay in deep Q-networks (DQN) [1] has greatly contributed to the development of reinforcement learning. In the past several years, these techniques have been successful with sequential decision tasks such as robot control [2], natural language processing (NLP) [3], autonomous vehicle decision-making [4,5], etc. However, most RL algorithms are still in their infancy and have very limited real-world applications.…”