Comparative Analysis of Trajectory Planning Algorithms in Semi-Structured Environments Using Reinforcement Learning.

• Grant number: 25/17147-1
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2026
• End date: January 31, 2027
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Denis Fernando Wolf
• Grantee: Vinicius Gustierrez Neves
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

This project aims to develop and evaluate trajectory planning algorithms for autonomous vehicles operating in semi-structured environments by leveraging Deep Reinforcement Learning (DRL) techniques. The research seeks to compare the performance of these approaches with traditional methods, such as the graph-based D* algorithm and Model Predictive Control, with an emphasis on dynamic and highly unpredictable scenarios. Validation will be conducted across four distinct types of scenarios, using metrics related to safety, comfort, and the efficiency of the generated trajectories. The CARLA simulator will be employed to build controlled environments, enabling systematic experimentation and the collection of reliable data. The expected results are to broaden knowledge about autonomous navigation strategies in partially structured contexts, offering more efficient solutions and a solid comparative analysis among the principal algorithms discussed in the literature. (AU)