Search

In farming systems, harvesting operations are tedious, time- and resource-consuming tasks. Deploying a fleet of autonomous robots to work alongside farmworkers may provide vast productivity and logistics benefits. In this context, an intelligent robotic system should monitor human behavior, identify the ongoing activities and anticipate the worker's needs. Unlike other application areas, such as warehouses and factories, research on human behavior recognition in agriculture is still in its infancy and has few case studies. Thus, there is a need for developing a fully integrated human activity recognition (HAR) methodology applied for agricultural operations in production fields. In this work, the main contribution consists of creating a benchmark framework of video-based human pickers detection, classifying their activities and corresponding motion direction, to serve in harvesting operations in different agricultural scenarios. Our solution uses the combination of a Mask Region-based Convolutional Neural Network (Mask R-CNN) for object detection and optical flow for motion estimation with a newly added statistical attribute of flow motion descriptors, named as Correlation Sensitivity (CS). A classification criterion is defined based on the analysis of the Kernel Density Estimation (KDE) technique and the K-Means clustering algorithm. Both methods are evaluated upon in-house collected datasets from different environments like strawberry polytunnels and apple tree orchards. The proposed benchmark framework is quantitatively analyzed using measures of sensitivity, specificity, and accuracy and shows satisfactory results amidst various dataset challenges such as multi-foreground objects, lighting variation, blur, and occlusions.

The development and integration of Wire Arc Additive Manufacturing (WAAM) systems is nowadays a topic of growing interest. Industries are starting to focus on deploying this technology due to its vast capability of producing different types of parts and creating new possibilities for engineering design. Measuring the process characteristics is crucial in a WAAM system, because it helps to ensure that the build up was done according as planned. In this work, it was developed a passive vision-based monitoring method to measure the metal bead width in a WAAM process. The deposition was carried out by using a carbon steel wire with GMAW process, a Motoman HP20 robot arm and a welding torch device. A Xiris XVC-1000camera was used for visual inspection and mounted in a suitable configuration to minimize arc's noise. The experimental results show that it is possible to measure the bead deposition width in real time with a satisfactory accuracy using monocular cameras. Therefore, it is a feasible solution to be used in WAAM systems with the advantage of being relatively low-cost, as compared to other active vision equipment.

This paper presents a novel 3-DoF parallel robot arm designed for use in applications that demand a high number of densely mounted manipulators with overlapping workspaces. The presented arm, which is mounted on a single rail, has such a workspace that allows the mechanism to reach outside the transverse footprint of its rail. Neighboring arms may therefore have overlapping workspaces on both sides and can be placed very close together, a property that is usually not found in parallel robots. The workspace of the arm in the direction of the rail is only limited by the length of the rail itself, and can therefore be made arbitrarily long. This makes the arm a light-weight alternative to traditional gantry systems. The main motivation is to increase the number of arms that can be mounted on a restricted area compared to conventional robots. The arm is applied to the agricultural domain where we show that it is well suited for tasks like robotic weeding and harvesting, i.e., cases where a high number of arms working in parallel is beneficial and can increase efficiency of the operations substantially. A kinematic analysis of the arm in terms of workspace and singularities is presented.

In this work, it is considered an image based visual servoing control problem, for uncertain robot manipulators. Visual feedback is provided by a fixed monocular camera with uncertain parameters, for the purpose of tracking translational trajectories of a spherical target, both the trajectory on image plane and depth. Based on a cascade structure, the proposed adaptive visual servoing is combined with an adaptive motion control strategy. Stability and passivity properties are analyzed with the Lyapunov method. Simulation results illustrate and highlight performance of the proposed controller.

In this work, we consider a novel adaptive hybrid pose/force control strategy for uncertain robot manipulators capable of performing interaction tasks on poorly structured environments. A unique hybrid control law, based on an orientation-dependent term, is proposed to overcome the performance degradation of the feedback system due to the presence of uncertainties in the geometric parameters of the contact surfaces. A gradient-based adaptive law, which depends on the tracking error, is designed to deal with parametric uncertainties in the robot kinematics and the stiffness of the environment. In our solution, the effect of the uncertain robot dynamics is addressed by using an adaptive dynamic control based on a cascade control strategy. The Lyapunov stability theory and the passivity paradigm are employed to carry out the stability analysis of the overall closed-loop control system. Numerical simulations are included to illustrate the performance and feasibility of the proposed methodology.

In recent years, the use of robots in agriculture has been increasing mainly due to the high demand of productivity, precision and efficiency, which follow the climate change effects and world population growth. Unlike conventional agriculture, sugarcane farms are usually regions with dense vegetation, gigantic areas, and subjected to extreme weather conditions, such as intense heat, moisture and rain. TIBA - Tankette for Intelligent BioEnergy Agriculture - is the first result of an R&D project which strives to develop an autonomous mobile robotic system for carrying out a number of agricultural tasks in sugarcane fields. The proposed concept consists of a semi-autonomous, low-cost, dust and waterproof tankette-type vehicle, capable of infiltrating dense vegetation in plantation tunnels and carry several sensing systems, in order to perform mapping of hard-to-access areas and collecting samples. This paper presents an overview of the robot mechanical design, the embedded electronics and software architecture, and the construction of a first prototype. Preliminary results obtained in field tests validate the proposed conceptual design and bring about several challenges and potential applications for robot autonomous navigation, as well as to build a new prototype with additional functionality., 9 pages, 15 figures

In this work, we consider the 3D visual tracking problem for a robot manipulator with uncertainties in the kinematic and dynamic models. The visual feedback is provided by a fixed and uncalibrated camera located above the robot workspace. The Cartesian motion of the robot end effector can be separated into a 1D motion parallel to the optical axis of the camera and a 2D motion constrained on a plane orthogonal to this axis. Thus, the control design can be simplified, and the overall visual servoing system can be partitioned in two almost-independent subsystems. Adaptive visual servoing schemes, based on a kinematic approach, are developed for image-based look-and-move systems allowing for both depth and planar tracking of a reference trajectory, without using image velocity and depth measurements. In order to include the uncertain robot kinematics and dynamics in the presented solution, we develop a cascade control strategy based on an indirect/direct adaptive method. The stability and convergence properties are analyzed in terms of Lyapunov-like functions and the passivity-based formalism. Numerical simulations including hardware-in-the-loop results, obtained with a robot manipulator and a web camera, are presented to illustrate the performance and feasibility of the proposed control scheme. Copyright © 2016 John Wiley & Sons, Ltd.

Neste trabalho, apresenta-se uma metodologia para a adaptacao tecnologica de um automodelo comercial para pesquisas basicas e avancadas na area de robotica agricola. A modelagem matematica do veiculo e baseada em um modelo cinematico simples do tipo biciclo, enquanto que o algoritmo de controle e projetado a partir de uma abordagem cinematica. Uma ideia basica consiste em adaptar a lei de controle desenvolvida para o modelo do uniciclo e, de acordo com algumas restricoes cinematicas, aplica-la para o biciclo. Simulacoes computacionais em Rviz, o ambiente de visualizacao 3D da plataforma ROS, sao realizadas para verificar e validar as abordagens de modelagem e controle apresentadas. Resultados experimentais sao incluidos para ilustrar a eficiencia e a viabilidade da metodologia proposta.

This work presents a methodology for autonomous manipulation of valves using a dual-arm robot and the visual sensing provided by cameras mounted on the head and each robot arm. A novel image-based identification and pose estimation method is devised to determine the valve to be manipulated as well as to estimate its position and orientation with respect to the robot base. The valve pose is used in the positioning and alignment of the robot end effector enabling the manipulation task to be carried out autonomously. Experimental results, obtained with a BaxterTM robot performing valve turning tasks, illustrate the feasibility of the proposed methodology. The main objective of this work is to provide capabilities to develop new solutions and innovative technologies to deal with the challenge of manipulating valves autonomously in poorly structured and harsh environments

In this work, we propose a solution to the inverse kinematics problem based on the differential kinematics and on a recently proposed algorithm which estimates the inverse of the Jacobian matrix dynamically. The output of the algorithm can be interpreted as a filtered inverse (FI) of the Jacobian matrix. An interesting property of the FI algorithm is its ability to cope with kinematic singularities. The update law of the estimator is driven by error signals that consider both the left and the right inverse matrices, thus enabling trajectory tracking and minimization of the control effort simultaneously. This paper shows that the FI algorithm can be applied to a Jacobian matrix augmented with additional constraints, which allows for setting the priority or weight to different control objectives, such as obstacle avoidance. Simulation results are presented to illustrate the performance and feasibility of the proposed solution.

El control servovisual de robots manipuladores ha sido un tema en evolución en los últimos años, especialmente en aplicaciones donde el ambiente es no estructurado o el acceso para un operador es difícil.Para diseñar este tipo de controladores se hace importante realizar simulaciones previas para ajustar los parámetros o tener una aproximación del comportamiento. En este trabajo se presentan dos diferentes modelos de sistemas de visión,enfocados a aplicaciones en el control de robots manipuladores. El modelado de las cámaras de video se obtendrá con la ya conocida proyección de perspectiva. Para validar los modelos se presentaran simulaciones de dos controladores servovisuales en 2D y 3D.

In this work the problem of remote calibration and trajectory replanning for a subsea robotic manipulator is considered. Because the trajectory planning is normally done in a structured environment, several uncertainties arise when the robot is placed on the seabed and these need to be compensated for to guarantee that the task specifications are fulfilled. We address the particular problem of configuration errors in the robot base with respect to the configuration used during the off-line trajectory planning. A calibration method based on both internal and external sensors is presented in order to estimate the uncertainty in the location of the robot. Moreover, a trajectory replanning strategy in the Cartesian velocity space is proposed to guarantee that the originally planned trajectory is followed. Simulation and experiments performed with a 6-DoF robot manipulator and a real calibration grid show the viability of the proposed planning and control schemes.

This paper addresses the posture control problem for robotic systems subject to kinematic constraints. The key idea is to consider the kinematic constraints of the mechanisms from their structure equations, instead of explicitly using the constraint equations. A case study for parallel robots and cooperating redundant robots is discussed based on the following concepts: forward kinematics, differential kinematics, singularities and kinematic control. Simulations results, obtained with a Four-Bar linkage mechanism, a planar Gough-Stewart platform and two cooperating robots, illustrate the applicability and versatility of the proposed methodology. Este artigo considera o problema de controle de postura para sistemas robóticos com restrições cinemáticas. A ideia principal é considerar as restrições cinemáticas dos mecanismos a partir de suas equações estruturais, ao invés de usar explicitamente a equação de restrição. Um estudo de caso para robôs paralelos e robôs cooperativos é discutido baseado nos conceitos de cinemática direta, cinemática diferencial, singularidades e controle cinemático. Resultados de simulação, obtidos a partir de um mecanismo Four-Bar linkage, uma plataforma de Gough-Stewart planar e dois robôs cooperativos, ilustram a aplicabilidade e versatilidade da metodologia proposta.

This work addresses the 3D visual servoing problem for robot manipulators in the presence of uncertainties in the camera-robot system parameters. From a fixed and uncalibrated camera, the cartesian motion is decomposed into a 2D motion on a plane orthogonal to the optical axis and a 1D motion parallel to this axis. An adaptive visual servoing scheme based on a kinematic approach is developed for image-based look-and-move systems to allow both depth and planar tracking of a reference trajectory, without using image velocity and depth measurements. A cascade control strategy based on direct/indirect adaptive method is applied to consider the uncertain robot kinematics and dynamics in the presented solution. The stability properties and the convergence analysis are discussed from the Lyapunov-passivity formalism. Simulation results are shown to illustrate the performance of the proposed control scheme.

This paper presents a control strategy for robot manipulators to perform 3D cartesian tracking using visual servoing. Considering a fixed camera, the 3D cartesian motion is decomposed in a 2D motion on a plane orthogonal to the optical axis and a 1D motion parallel to this axis. An image-based visual servoing approach is used to deal with the nonlinear control problem generated by the depth variation without requiring direct depth estimation. Due to the lack of camera calibration, an adaptive control method is used to ensure both depth and planar tracking in the image frame. The depth feedback loop is closed by measuring the image area of a target object attached to the robot end-effector. Simulation and experimental results obtained with a real robot manipulator illustrate the viability of the proposed scheme. Este trabalho apresenta uma estratégia de controle para robôs manipuladores realizarem rastreamento cartesiano 3D utilizando servovisão. Considerando uma câmera fixa, o movimento cartesiano 3D é decomposto em um movimento 2D sobre um plano ortogonal ao eixo óptico e em outro movimento 1D paralelo ao mesmo eixo. Uma abordagem de servovisão baseada em imagem é utilizada para tratar o problema de controle não-linear, gerado pela variação de profundidade, sem a necessidade de estimar esta medida. Devido à ausência de calibração da câmera, um método de controle adaptativo é utilizado para assegurar rastreamento planar e de profundidade nas coordenadas da imagem. A malha de controle de profundidade é fechada através da medição da área da imagem de um objeto fixado no efetuador do robô. Simulação e resultados experimentais, obtidos com um robô manipulador real, ilustram a viabilidade do esquema proposto.

Neste trabalho, considera-se um sistema robótico servovisual usando câmera fixa e não-calibrada. Dois controladores adaptativos são propostos para rastrear visualmente um alvo em movimento na presença de incertezas nos parâmetros da câmera. A primeira estratégia adaptativa utiliza o método de Hierarquia de Controle para solucionar o problema de controle adaptativo multivariável. A segunda é baseada na simetrização via fatoração da matriz de controle. Os algoritmos adaptativos são robustos na medida em que apresentam reduzida sensibilidade à incertezas cinemáticas. Resultados experimentais com um manipulador em movimento planar ilustram a robustez e a viabilidade dos esquemas propostos. In this work, a robotic visual servoing system using a fixed and uncalibrated camera is considered. Two adaptive controllers are proposed for visually tracking a moving target in the presence of camera parameters uncertainties. The first adaptive strategy uses the Hierarchy of Control method to solve the multivariable adaptive control problem. The second one is based on the symmetrization via factorization of the control matrix. The adaptive algorithms are robust in the sense that they present reduced sensitivity to kinematic uncertainties. Experimental results with a manipulator in a planar motion illustrate the robustness and viability of the proposed schemes.