Alexandre C. B. Ramos

Application of Self-Organizing Maps at Change Detection in Amazon Forest

This paper presents a change detection of aerial images algorithm based on Kohonen self-organizing maps. In this study, a set of images from different period of time from Tapajós National Forest at Amazon forest in Pará, Brazil, has been employed. As input of the self-organizing maps, used as an unsupervised neural network, it was obtained features from the images using three different size of square windows and as output, a map of 16 colors, as 16 groupings. The results are obtained through the comparison between the outputs obtained with these different sizes of windows.

Dynamic Simulation of the Flight Behavior of a Rotary-Wing Aircraft

In this paper we seek to present the mathematical model relating to flight dynamics of a rotary-wing aircraft. The procedure starts by linearizing the translational and rotational dynamics and rotational kinematic equations of motion based on perturbation theory. Some procedures are simplified and are implemented here, due to the complex modeling process. The second step is to obtain the linear form, which is fundamental in order to describe the stability and response of the small movements of the helicopter around a specific attitude. Finally, the dynamic behavior of two helicopters (AS355-F2 Squirrel and BO105-S123) was simulated employing MATLAB Simulink v7.6 (2008) [1]. The input data, status and control derivatives from an existing helicopter were employed. It is noteworthy that the linearized model developed here is valid for applications where a deeper representation of the aircraft is not required.

Low cost helicopter training simulator a case study from the brazilian Military Police

This work describes a computer based training system to assist the crewmembers to learn the ground school of helicopter AS350-B2, manufactured from HELIBRAS, a brazilian helicopter company and widely used by the Armed Forces, Civil and Military Policies and beyond companies of executives transport and load transport. The training system consists of a 1∶1 scale aircraft cabin simulator, a tutoring system software and a MS Flight Simulator interface. The project is in advanced phase of development and already it possesss positive depositions of responsible interested parties for the sectors of training of the companies and institutions that operate the aircraft. A case study from brazilian Military Police is presented.

Neural Inversion of Flight Guidance Dynamics

Differential flatness, a property of some dynamic systems which has been recognized only recently, has made possible the development of new tools to control complex nonlinear dynamic systems. Many dynamic nonlinear systems have been proved to be differentially flat. In this paper, it is shown that the inertial position coordinates of an aircraft can be considered as differential flat outputs for its flight guidance dynamics. Since this differential flatness property is implicit, a neural network is introduced, as a numerical device, to deal with the inversion of the guidance dynamics. Then, once conveniently structured and trained, a neural network is able to generate in real time directives to conventional autopilot systems so that a reference trajectory can be tracked. Numerical results relative to the training of the neural network and to trajectory tracking are displayed and discussed.

Change Detection in Satellite Images Using Self-Organizing Maps

In several applications it is necessary to detect changes from aerial images analysis. The goal, which has been addressed in several studies, is to develop an automatic method with low computational cost. However, it is a difficult task and isolated contributions have been presented. The main purpose of this paper is present a proposal to detecting change using aerial images obtained by satellite. Kohonens Self-Organizing Maps (SOM) is used to identify the classes of the images and the change detection is based on post-classification technique. In this paper, a set of images from different period of time from Brazil has been employed. As input of the self-organizing maps, used as an usupervised and nonparametric artificial neural network, it was obtained different features. The self-organizing map is evaluated by varying the type of topological neighborhood function, as a Gaussian, truncated and square Gaussian. The results are obtained through the comparison between the outputs obtained with these maps of colors. The experiments performed on the satellite image have shown that SOM is efficient and have better results for the area of study were obtained using Gaussian and Truncated Gaussian functions.

Using ANN and UAV for terrain surveillance

Autonomous Unmanned Aerial Vehicles (UAVs) provide an effective alternative for surveillance in urban areas due to their cost and safety when compared to other traditional methods. The objective of this study is to report the development of a system capable of analyzing digital images of the terrain and identifying potential invasion, unauthorized changes in land and deforestation in some special urban areas. Images are captured by a camera attached to an autonomous helicopter, flying it around the area. For processing the images, an Artificial Neural Network (ANN) technique called Self Organizing Map (SOM) is used.

Expanding small UAV capabilities with ANN: A case study for urban areas observation

Autonomous Unmanned Aerial Vehicles (UAVs) are available alternatives for urban areas inspections due to its cost and safety when compared to other traditional methods. The purpose of this paper is to report the development of a system capable of analyzing digital images of the terrain and identifies potential invasion, unauthorized alterations on the ground and deforestation in some areas of special use. Images are captured by a camera coupled to an autonomous helicopter, which flight around the area. For the processing of the images an artificial neural network technique called Kohonen SOM (Self Organized Map) will be used. The processing is actually a set of steps that seek to collate the final common characteristics of a given image.

Neural Networks for an Analysis of the Hemometabolites Biosensor Response

In this work, the concentration dependent response of amperometric biosensor array for the biomarkers glucose, cholesterol and urease was explored, using artificial neural nets ANN. The aim was to explore an array of amperometric biosensors for the discrimination of the biomarkers glucose, cholesterol and urea in blood. Seven out of eight platinum electrodes on the array were modified with four different enzymes; glucose oxidase, cholesterol, urease and peroxidase. The dynamic biosensor response curves from the eight sensors were used for ANN analysis. The ANN were applied to an analysis of the biosensor response to multi-biomarkers mixtures the ANN was able to detect the conditions with an accuracy up to 90%. The results obtained by using ANN to interpret the electrical signal of the developed biosensor arrays leads to the conclusion that: i after training the ANN, the evaluation of recorded data are on-line, ii microelectrode sites which are highly correlated to the information about the concentrations within the recorded signals was identified, iii the recognition of blood biomarkers is improved by using the ANN.

Modeling of System Software for Computer Based Training

With major technological advances and to reduce the cost of training apprentices for real-time critical systems, it was necessary to the development of Intelligent Tutoring Systems for training apprentices in these systems. These systems, in general, have interactive features so that the learning is actually more efficient, making the learner is more familiar with the mechanism in question. In the home stage of learning, tests are performed to obtain the students income, a measure on their use. The aim of this paper is to present the modeling of Tutorial of Intelligent Systems using the UML language. The various steps of the analysis are considered the diagrams required to build a general model, whose purpose is to present the different perspectives of its development.

Collision Avoidance Based on Reynolds Rules: A Case Study Using Quadrotors

This work aims to present a collision avoidance algorithm developed to drive a swarm of Unmanned Aerial Vehicles (UAVs) using Reynolds flocking rules. We used small quadrotor aircrafts with 250 mm diameter, equipped with GPS and distance sensors, controlled by a Pixhawk autopilot board and an embedded Linux computer (Raspberry Pi). The control algorithm was implemented in C++ as a package for the ROS platform and runs on the Raspberry Pi, while the Pixhawk is responsible for the low level control of the aircraft. The distance sensors are used to detect nearby robots and based on this information the control algorithm determines the direction the robot should move to avoid collisions with its neighbors. We are able to see that from the individual perception and local interaction of each robot a group behavior emerges and the swarm move together coherently. A simulation environment based on the Gazebo Simulator was prepared to test and evaluate the algorithm in a way as close to the reality as possible.