Decision Fault Tree Learning and Differential Lyapunov Optimal Control for Path Tracking

Abstract

This paper considers the main challenges for all components engaged in the driving task suggested by the automation of road vehicles or autonomous cars. Numerous autonomous vehicle developers often invest an important amount of time and effort in fine-tuning and measuring the route tracking to obtain reliable tracking performance over a wide range of autonomous vehicle speed and road curvature diversities. However, a number of automated vehicles were not considered for fault-tolerant trajectory tracking methods. Motivated by this, the current research study of the Differential Lyapunov Stochastic and Decision Defect Tree Learning (DLS-DFTL) method is proposed to handle fault detection and course tracking for autonomous vehicle problems. Initially, Differential Lyapunov Stochastic Optimal Control (SOC) with customizable Z-matrices is to precisely design the path tracking for a particular target vehicle while successfully managing the noise and fault issues that arise from the localization and path planning. With the autonomous vehicle’s low ceilings, a recommendation trajectory generation model is created to support such a safety justification. Then, to detect an unexpected deviation caused by a fault, a fault detection technique known as Decision Fault Tree Learning (DFTL) is built. The DLS-DFTL method can be used to find and locate problems in expansive, intricate communication networks. We conducted various tests and showed the applicability of DFTL. By offering some analysis of the experimental outcomes, the suggested method produces significant accuracy. In addition to a thorough study that compares the results to state-of-the-art techniques, simulation was also used to quantify the rate and time of defect detection. The experimental result shows that the proposed DLS-DFTL enhances the fault detection rate (38%), reduces the loss rate (14%), and has a faster fault detection time (24%) than the state of art methods.

Comparing Probabilistic Logic Factored MDPs, CART and MLPs for Behavior Selection in Self-Driving Cars

“…In (Spanogiannopoulos et al, 2022) authors propose generating real-time, collision-free paths in unknown but near static environments for non-holonomic self-driving cars, using an incremental planner based on rapidly-exploring random trees. The study stated in (Bose et al, 2023) explores challenges in automating road vehicles, with an emphasis on fault-tolerant trajectory tracking. It introduces the differential Lyapunov stochastic and decision fault tree learning for precise and efficient path tracking amidst noise and faults.…”

Hessian Distributed Ant Optimized Perron–Frobenius Eigen Centrality for Social Networks

“…A decomposition technique for multi-project scheduling that is based on two stages and has a suitable CPU running time for large-size instances was devised in [12]. To identify deliberate activities on highways, a multi-agent system for smart roads was created in [13] utilizing Model Driven Engineering (MDE). For decentralized multi-agent learning addressing high dimensions, a traditional rewarding system in gaming notion utilizing a Deep Q-learning framework was presented in [14].…”