carolinarutililima

ADA_Santander

patrolDES

Multi-robot systems that interact with non-deterministic environments, in addition to cooperating to achieve global objectives efficiently, each robot must be able to react safely and without blocking to uncertain environmental and hardware events. This work applies the Supervisory Control Theory (SCT) to a multi-robot system for region patrolling. Plant and specification modeling and modular synthesis of supervisors for a small multi-robot patrolling problem subject to equipment failures and battery uncertainties are presented. A deliberative/reactive SCT-based hybrid architecture for distributed deployment in mobile robots is proposed, where safety and non-blocking requirements are met by optimal supervisors, while heuristics exploit the flexibility of supervisory control to make patrolling efficient. For architectural validation purposes, a prototype was implemented in a simulation environment, under which several experiments were performed comparing four different control strategies. The results indicate that the deliberative/ reactive architecture is viable and robust for real robots, allowing to ensure safety and efficiency in multi-robot patrols even in the presence of unexpected events.

Supervised-Learning-Algorithms

Trabalho_ADA_ProgII

KNN and K-Means from scratch using a dataset of images from https://www.kaggle.com/c/dogs-vs-cats/data

Trabalho_Final_Tecnicas_Prg

auction_ag

Bongo_Board

It presents the performance of the Police Gradient reinforcement learning algorithm using the "Bong Board" as environment. The Bong Board is a linear inverted pendulum.

carolinarutililima.github.io

Dimension-Reduction-Techniques

Dimension-Reduction-Techniques_2

carolinarutililima/carolinarutililima.github.io

carolinarutililima/Tree_Learning

Projeto final de matétia de Matemática da Computação

carolinarutililima/pdf_to_csv

carolinarutililima/ADA_Santander

carolinarutililima/Final_Project_Machine_Learning_II

carolinarutililima/MPC_Casadi

carolinarutililima/Final_Project_Machine_Learning_I

carolinarutililima/Projeto_Final_Estatistica

Projeto final do curso de Estatística I da ADA

carolinarutililima/SQL_Final_Project

Projeto final do curso de SQL da ADA

carolinarutililima/Trabalho_Final_Tecnicas_Prg

carolinarutililima/Trabalho_ADA_ProgII

KNN and K-Means from scratch using a dataset of images from https://www.kaggle.com/c/dogs-vs-cats/data

carolinarutililima/image_maskrcnn

In this research we used multiple dataset images to detect carcerigenous cell in whole tissues.

carolinarutililima/Unsupervised-Learning-Algorithms

carolinarutililima/Supervised-Learning-Algorithms

carolinarutililima/Dimension-Reduction-Techniques_2

carolinarutililima/Dimension-Reduction-Techniques

carolinarutililima/Multivariate-Statistical-Inference

carolinarutililima/Regression2

carolinarutililima/Regression-Analisys

carolinarutililima/olist-ml-models

Projeto de Machine Learning do início ao fim no contexto de um e-commerce

carolinarutililima/mask_rcnn_mod

carolinarutililima/letsmove

A simple game using Google's MoveNet, also the only example for a working webcam and Phaser 3 I think.

carolinarutililima/Kinematics_Robot

carolinarutililima/walking_gait_robot

carolinarutililima/NSGA

carolinarutililima/simulated-annealing

carolinarutililima/pso

particle swarm optimization (PSO) applied to Ackey and quadratic function.

carolinarutililima/GA

This assignment approaches the Genetic Algorithm performance from two benchmark functions, quadratic and Ackley. Moreover, it will be presented a performance and comparison changing some parameters in the genetic algorithm. Genetic Algorithms are used to find approximate solutions to search and optimization problems. Furthermore, this kind of algorithm is inspired by Charles Darwin's theory of natural evolution, i.e., this algorithm reflects the process of natural selection where the fittest individuals are selected for reproduction to produce offspring of the next generation.

carolinarutililima/patrolDES

Multi-robot systems that interact with non-deterministic environments, in addition to cooperating to achieve global objectives efficiently, each robot must be able to react safely and without blocking to uncertain environmental and hardware events. This work applies the Supervisory Control Theory (SCT) to a multi-robot system for region patrolling. Plant and specification modeling and modular synthesis of supervisors for a small multi-robot patrolling problem subject to equipment failures and battery uncertainties are presented. A deliberative/reactive SCT-based hybrid architecture for distributed deployment in mobile robots is proposed, where safety and non-blocking requirements are met by optimal supervisors, while heuristics exploit the flexibility of supervisory control to make patrolling efficient. For architectural validation purposes, a prototype was implemented in a simulation environment, under which several experiments were performed comparing four different control strategies. The results indicate that the deliberative/ reactive architecture is viable and robust for real robots, allowing to ensure safety and efficiency in multi-robot patrols even in the presence of unexpected events.

carolinarutililima/Bongo_Board

It presents the performance of the Police Gradient reinforcement learning algorithm using the "Bong Board" as environment. The Bong Board is a linear inverted pendulum.