Human–Robot Collaboration is a critical component of Industry 4.0, contributing to a transition towards more flexible production systems that are quickly adjustable to changing production ...requirements. This paper aims to increase the natural collaboration level of a robotic engine assembly station by proposing a cognitive system powered by computer vision and deep learning to interpret implicit communication cues of the operator. The proposed system, which is based on a residual convolutional neural network with 34 layers and a long-short term memory recurrent neural network (ResNet-34 + LSTM), obtains assembly context through action recognition of the tasks performed by the operator. The assembly context was then integrated in a collaborative assembly plan capable of autonomously commanding the robot tasks. The proposed model showed a great performance, achieving an accuracy of 96.65% and a temporal mean intersection over union (mIoU) of 94.11% for the action recognition of the considered assembly. Moreover, a task-oriented evaluation showed that the proposed cognitive system was able to leverage the performed human action recognition to command the adequate robot actions with near-perfect accuracy. As such, the proposed system was considered as successful at increasing the natural collaboration level of the considered assembly station.
•Implicit communication cues are crucial for effortless Human–Robot Collaboration.•Human actions contain task-focused information that provide operation context.•A Deep Learning cognitive system was proposed to interpret task-focused human actions.•The proposed cognitive system was applied in a real collaborative assembly scenario.•Great accuracy was achieved at recognizing human actions and commanding robot tasks.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
The trends of the modern industrial sector indicate a transition to more flexible production systems that are quickly adjustable to changing production requirements. Human-robot collaboration is one ...of the central components of this paradigm shift, which was brought forward by the technological breakthroughs of Industry 4.0 and is seen as critical to increase enterprise competitiveness, particularly in developed countries.With these ideas in mind, the current dissertation aims to increase the natural collaboration level of a robotic engine assembly station by developing a cognitive system powered by computer vision and deep learning to interpret implicit communication cues of the operator. Such system should focus on obtaining assembly context through action recognition of the tasks performed by the operator. This will provide the robot with the ability to respond to the actions of the operator without the need of being prompted to do so, thus improving the collaboration level of the assembly station.A dedicated dataset was proposed and recorded for the considered collaborative engine assembly task, allowing for deep learning models to be trained for the use case at hand. Due to limitations of availability of human resources, only one operator participated in the recording, compromising the ability of the trained models to work with any operator without prior adjustments. Nevertheless, solutions were proposed to surpass this issue, such as the recording of a more variable dataset or the implementation of transfer learning to fine-tune the proposed model. Regarding the cognitive system itself, several deep learning model architectures were proposed and tested to perform the intended human action recognition, including a 3D convolutional neural network, and two different variations of 2D convolutional neural networks and recurrent neural networks. The selected model architecture for the final system was a residual convolutional neural network with 34 layers and a long-short term memory residual neural network (ResNet-34 + LSTM).This model showed a great performance, achieving an accuracy of 96.65% and a temporal mean intersection over union (mIoU) of 94.11%. Moreover, a task-oriented evaluation showed that the proposed cognitive system was able to leverage the performed human action recognition to command the adequate robot actions with near-perfect accuracy. As such, the proposed system was considered as successful at increasing the natural collaboration level of the considered assembly station.
Muitas indústrias modernas ainda dependem dos processos tradicionais de assemblagem manual, com as suas típicas limitações de eficiência, qualidade inconsistente do produto final e, em alguns casos, ...problemas ergonómicos para os trabalhadores. A implementação de estações de assemblagem robotizada com ambiente de trabalho estruturado permite ultrapassar estas limitações, mas só podem ser utilizadas em indústrias de produção em massa devido aos seus elevados custos. Em contrapartida, a implementação de estações flexíveis de assemblagem robotizada necessita da utilização de soluções tecnológicas como controlo de força ou visão artificial para ultrapassar as incertezas posicionais presentes num ambiente de trabalho não estruturado. Com o objetivo de materializar uma estação móvel de assemblagem flexível, a presente dissertação procura desenvolver estratégias baseadas em controlo de força para ultrapassar as incertezas posicionais provenientes do sistema de posicionamento da plataforma móvel.As capacidades de controlo de força do manipulador utilizado (KUKA LBR iiwa 14R 820) foram exploradas, culminando em duas estratégias de assemblagem distintas. Métodos baseados em estratégias de oscilação são capazes de realizar montagens simples, através da execução de rotinas de “procura cega” de modo a inserir uma determinada peça na correspondente ranhura. Este método não elimina as incertezas presentes no ambiente de trabalho e, como tal, tem de ser executado individualmente para cada assemblagem realizada. Por outro lado, o método baseado em exploração do ambiente complementado por correção de orientação utiliza movimentos para sondar o ambiente de trabalho e eliminar completamente as incertezas presentes. Por esta razão, este método é adequado para estações com várias montagens, uma vez que a rotina só necessita de ser executada uma vez.Como a implementação do método de exploração do ambiente é dependente da geometria da estação de assemblagem, este método não pode ser diretamente utilizado noutras estações de montagem. Assim, com o objetivo de ultrapassar esta limitação e generalizar a rotina desenvolvida para qualquer operação de assemblagem, foi proposto um acessório externo, composto por duas superfícies ortogonais. Este acessório permite a implementação otimizada deste método em qualquer assemblagem com incertezas posicionais planares. Alternativamente, a geometria de cada estação de montagem pode ser analisada de modo a tentar identificar duas superfícies ortogonais desimpedidas, nas quais os movimentos de sonda possam ser realizados.A implementação das rotinas desenvolvidas na configuração final do manipulador móvel permitiu analisar o impacto relativo de cada um dos erros de posicionamento da plataforma. Foi concluído que os erros de orientação da plataforma móvel são diretamente responsáveis por grandes erros de translação na montagem e, como tal, devem ser limitados ao máximo. Além disso, como esta correspondência de erros é amplificada pela distância entre o manipulador e o local da montagem, esta distância também deve ser tanto menor quanto possível.
Quantum walks provide a natural framework to approach graph problems with quantum computers, exhibiting speedups over their classical counterparts for tasks such as the search for marked nodes or the ...prediction of missing links. Continuous-time quantum walk algorithms assume that we can simulate the dynamics of quantum systems where the Hamiltonian is given by the adjacency matrix of the graph. It is known that such can be simulated efficiently if the underlying graph is row-sparse and efficiently row-computable. While this is sufficient for many applications, it limits the applicability for this class of algorithms to study real world complex networks, which, among other properties, are characterized by the existence of a few densely connected nodes, called hubs. In other words, complex networks are typically not row-sparse, even though the average connectivity over all nodes can be very small. In this work, we extend the state-of-the-art results on quantum simulation to graphs that contain a small number of hubs, but that are otherwise sparse. Hopefully, our results may lead to new applications of quantum computing to network science.
This paper presents a collaborative mobile manipulator assembly station, which uses force control to surpass the positional uncertainties arising from unstructured work environments and positional ...errors of the mobile platform. For this purpose, the use case of an internal combustion engine for the automotive industry was considered. Several force control heuristics relying on blind searches using oscillations and/or environment exploration were developed and implemented. Particular attention was given to the orientation errors of the mobile platform, as it was proved that they have a significant impact on the assembly task. The proposed heuristics showed great potential for the use case at hand. Particularly, when the orientation error of the platform is limited to ±2°, the oscillation method complemented by environment exploration was able to surpass a maximum translation error of 32.3mm, whereas the environment exploration complemented by orientation correction was able to surpass an error of 73.3mm. Moreover, a generalization strategy was proposed, intending to expand the scope of the developed heuristics to other assembly applications.
A investigação acerca de um jornal feminista, cujas colaboradoras são contemporâneas da geração masculina dos anos setenta, afigurou- -se-nos de interesse no âmbito dos estudos sobre as mulheres.A ...exploração minuciosa desta publicação permitiu-nos colher informações sobre uma multiplicidade de temas e obter uma visão da Europa e do mundo filtrada por uma óptica feminista, na medida em que interessava evidenciar todos os acontecimentos que contribuíssem para demonstrar a existência de mutações que incluíam as mulheres.Havia ainda uma intenção pedagógica, que pretendia apontar os erros e conduzir a modificações de mentalidades e comportamentos por parte das leitoras. Ficamos assim a conhecer mais profundamente a esfera em que se movimentavam as mulheres e a complexa problemática envolvente. Numa época em que predominava o aforismo "la femme c'est la maison", importava indicar muito claramente como desempenhar esse papel, que exigia conhecimentos.
Link prediction methods use patterns in known network data to infer which connections may be missing. Previous work has shown that continuous-time quantum walks can be used to represent path-based ...link prediction, which we further study here to develop a more optimized quantum algorithm. Using a sampling framework for link prediction, we analyze the query access to the input network required to produce a certain number of prediction samples. Considering both well-known classical path-based algorithms using powers of the adjacency matrix as well as our proposed quantum algorithm for path-based link prediction, we argue that there is a polynomial quantum advantage on the dependence on N, the number of nodes in the network. We further argue that the complexity of our algorithm, although sub-linear in N, is limited by the complexity of performing a quantum simulation of the network's adjacency matrix, which may prove to be an important problem in the development of quantum algorithms for network science in general.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
A Igreja e o Estado - Duarte Silva, Isabel Costa - Fernando Moutinho - Odeon; Disco Odeon, XO497, 143032; 143032 - access copy of digitized shellac recording (at 78rpm) in mp3 format