Robinson Jiménez Moreno

Monitoring system for global solar radiation, temperature, current and power for a photovoltaic system interconnected with the electricity distribution network in Bogota

This paper describes the implementation of a monitoring system to determine the efficiency of energy production generated by a set of six photovoltaic panels, interconnected with the domestic electricity grid in Bogota Colombia. The system is easy to use and enables the visualization of different signals acquired with sensors to measure temperature, solar radiation, voltage, DC current produced by the panel, the AC output of the inverter, the power generated by PV modules network and delivered power.

Trajectory planning for a robotic mobile using fuzzy c-means and machine vision

This article describes the development of a trajectory planning system which controls the movement of a tele-commanded mobile robot through machine vision techniques and fuzzy algorithms. Using a conventional camera, the scene in which the mobile is moving is captured. Then, by means of image processing techniques, the obstacles are segmented and using a variation of the fuzzy c-means algorithm, the possible paths to follow are delimited. The communication between the mobile and the central control system on a PC is done wirelessly via XBEE, so that the robot moves locally using a microcontroller. Using the developed algorithm it is possible to get a direct travel path without crashes, in which it is possible to vary the number of obstacles and where the traveling time is minimized.

Convolutional neural network architecture for hand gesture recognition

This paper presents the design of a convolutional neural network architecture using the MatConvNet library for MATLAB in order to achieve the recognition of 2 classes of hand gestures: ”open” and ”closed”. Six architectures were implemented to which their hyperparameters and depth were varied to observe their behavior through the validation error in the training and accuracy in the estimation of each one of the set classes, which was evaluated by a matrix of confusion. Given each of these results, the neural network with the best performance was chosen.

Design and construction of a prototype rehabilitation machine to hand and wrist

This paper presents the development of a machine for upper limb rehabilitation, focusing on the fingers and wrist, and is connected to a program with the voice command option which controls the exercise rehabilitation in addition to monitoring the patients progress.

Tracking of human operator arms oriented to the control of two robotic arms

This article presents the development of machine vision algorithms that allows to manipulate two robotic arms by the operator, replicating the movements of his own arms, accomplishing control of four degrees of freedom per arm. The development is aimed to provide an environment of natural, wireless and non-invasive control, providing security to the operator in the face of achievement of unsafe tasks, either for removal of hazardous, explosive or radioactive materials, as well as possibly toxic environments or just remote control from a remote site. The tracking algorithm of the two arms is based on their detection using the skeleton tool of Microsoft Kinect sensor, as well as image processing algorithms like color space conversion, segmentation and thresholding to control the gripper arm. It was possible to obtain real-time control of the arms to four degrees of freedom, according to the monitoring of the arms of the operator in the ranges of sensor coverage.

Robotic explorer to search people through face detection

This paper presents the implementation of an autonomous mobile robot which is capable of avoiding obstacles while it moves by means of an ultrasonic detection system and aims to find human faces. In a disaster such as a landslide or earthquake, where is possible that the initial condition of danger persists as would be the case of an earthquake, in which within a building, beams and parts of the structure remain committed, as well as the people who risk their lives finding victims. In response, the designed robot provides a solution by using artificial intelligence algorithms and image processing running on embedded systems. The obtained results allow the robot to move autonomously avoiding obstacles and stopping when it finds a human face.

Tree-Structured CNN for the Classification of Surgical Instruments

This paper presents the development of an algorithm of object classification focused on the detection and categorization of surgical instruments using tree-structured convolutional neural networks for the classification of initially 5 different tools and later evaluated with 10, managing to develop a new way of classifying objects capable of differentiating elements with a high resemblance to each other. Through partial division of the categories and kernel training focused on the most relevant changes of each instrument, percentages of about 70% accuracy were reached for a total of 10 elements, and 96% for 5 instruments.

Obstacle Evasion Algorithm for Clustering Tasks with Mobile Robot

This paper presents a proposal of obstacle evasion oriented to mobile robots in clustering tasks. For this case, polar coordinates are set for the movement of the mobile, the possible obstacles in the path are determined and imaginary boundaries are generated in each possible obstacle in order to delimit the path of the mobile between them. The algorithm developed under the Netlogo programming environment makes it possible to perform evasion and reach the clustering point efficiently.

Simulación del a reconversión energética de un vehículo de combustión a eléctrico con energía solar

En este articulo se presentan los resultados obtenidos en la reconversion energetica y el diseno aerodinamico de un prototipo de vehiculo unipersonal electrico

Planeación de trayectorias para un móvil robótico en un ambiente 3D

This article describes the development of a path generating system for the movement of a remote commanded robotic mobile, through machine vision techniques and fuzzy decision systems. Using a Kinect sensor, the system captures the motion area of the mobile and using image processing techniques, obstacles are segmented and a fuzzy clustering algorithm is implemented to define the possible paths to follow based on height and ability of mobile to exceed the obstacle. Communication between the mobile and the control system is performing wirelessly using a XBEE device which is centralized into PC software; therefore the robot travels locally through a microcontroller. By the developed algorithm is able to get a travel path with a minimum run time, in which it is possible to change the number of obstacles and approach paths with a slope up to 35 degrees of elevation from the ground.