Projects
This works presents a learning-based vehicle control system capable of navigating autonomously. Our approach is based on image processing, road and navigable area recognition, template matching classification for navigation control, and trajectory selection based on GPS way-points. The vehicle follows a trajectory defined by GPS points avoiding obstacles using a single monocular camera and maintaining the vehicle in the road lane. Different parts of the image, obtained from the camera, are classified into navigable and non-navigable regions of the environment using neural networks. They provide steering and velocity control to the vehicle. Several experimental tests have been carried out under environmental conditions to evaluate the proposed techniques.
A key challenge for the use of unmanned aerial vehicles (UAVs) is the security of their information during navigation to accomplish its task. Information security is a known issue, but it seems to be overlooked from a research perspective, that tends to focus on more classical and well-formulated problems. This paper addresses an experimental evaluation of three Denial of Service (DoS) attack tools to analyze the UAV's behavior. These tools are executed in real-time on the robot while it navigates an indor environment (inside the University building). We present experiments to demonstrate the impact of such attacks on a particular UAV model (AR.Drone 2.0) and also show a description of existing vulnerabilities. Our results indicate that DoS attacks might cause network availability issues influencing critical UAVs applications, such as the video streaming functionality.
The detection of failures (DF) in coffee crops is fundamental in evaluating product quality and the optimal occupation of planted areas. The use of unmanned aerial vehicles (UAVs) in precision agriculture has great potential as a tool to analyze critical parameters in cultivation, among them the detection of planting failures. This letter presents a novel methodology for DF from aerial images, obtained using a UAV capable of collecting high-resolution RGB images. The proposed uses mathematical morphology operators to detect failures over planted areas and returns both the individual positions of these failures and total failure length (sum of empty spaces between plants), facilitating decision making for further actions. Results show that proposed DF method is reliable for accurately identifying failures over rows of planted coffee crops.
Indoor navigation for mobile robots
Indoor navigation is a challenging task as many moving and movable objects can be found in the way, blocking the robot's path. In this project we aim to apply dynamic algorithms to replan paths more efficiently in case the environment changes. These algorithms must also allow robots to ask for human assistance to move objects on their surrounding in order to compute shorter paths.
This paper presents a mobile control system capable of learn behaviors based on human examples. Our approach is based on image processing, template matching, finite state machine, and template memory. The system proposed allows image segmentation using neural networks in order to identify navigable and non-navigable regions. It also uses supervised learning techniques which work with different levels of memory of the templates. As output our system is capable controlling speed and steering for autonomous mobile robot navigation. Experimental tests have been carried out to evaluate the learning techniques.
An UAV is used to obtain RGB images of a predefined area. The methodology enables the extraction of visual features from image regions and uses several Machine Learning (ML) techniques to classify regions into three classes: ground, healthy and diseased plants. Several ML techniques were compared using data obtained from a Eucalyptus crops. Results show that GP learning model is reliable than other ML for accurately identifying diseased trees.
A big challenge in the precision agriculture is the detection of fruits in coffee crops on agricultural environments. This paper presents a comparison of four features set to detect the red fruits (mature) in Coffee plants. An Unmanned Aerial Vehicle (UAV) is used to obtain high-resolution RGB images of a coffee hall. The proposed enables extraction of visual features from image regions and uses supervised Machine Learning (ML) techniques to classify areas as coffee fruits and non-fruits (branches and leaves). Several ML methods were compared using test data achieved from a Coffee plantation. Results show that ANN is reliable than other ML methods for accurately identifying coffee fruits.
Dev. and Assessment of a low-cost terrestrial mobile mapping system for transportation applications
This project evaluates a Terrestrial Mobile Mapping System (TMSS) based on the integration of low cost set of equipment. An open-source electronic prototype platform, shields GNSS, INS module, and five GoPro® cameras will be used. All items will be integrated via hardware/software, and assembled on a mobile platform. It is intended to conduct experiments in real conditions and assess the results – position quality index of extracted road system attributes (signs). The evaluation will be carried out by comparing the attributes 3D coordinates with a reference dataset that will be collected by conventional surveying (total station and high precision GNSS receiver). It is expected to describe applications for low cost TMMS based on experiments, also to produce an assembly material to be used for research groups or general users.
Multi-sensor localization for mobile robots within indoor environments
Although many algorithms and sensors have been proposed to indoor localization, none have proven to work perfectly under all situations. For instance, Global Positioning System (GPS) is not applied as signals can not penetrate most buildings, Star Gazer systems need marks distributed over the environment, depth cameras and laser range finder approaches are affected by moving and movable objects that change internal robot maps, and WiFi techniques have a minimum error greater than that of other approaches. In this project we combine different localization techniques using low-cost devices to improve robots localization accuracy.
Multi-agent coordination
Intelligent agents, such as service robots and self-driving cars, aim at helping humans in their daily tasks. Some of these tasks can be performed by a team of agents in order to reduce time and effort. This project focuses on the development of multi-agent coordination algorithms regarding the optimization of many objectives simultaneously.