Reward-modulated Spike-Timing-Dependent Plasticity (R-STDP) is a learning method for Spiking Neural Network (SNN) that makes use of an external learning signal to modulate the synaptic plasticity produced by Spike-Timing-Dependent Plasticity (STDP). Combining the advantages of reinforcement learning and the biological plausibility of STDP, online learning on SNN in real-world scenarios can be applied. This paper presents a fully digital architecture, implemented on an Field-Programmable Gate Array (FPGA), including the R-STDP learning mechanism in a SNN. The hardware results obtained are comparable to the software simulations results using the Brian2 simulator. The maximum error is of 0.083 when a 14-bits fix-point precision is used in realtime. The presented architecture shows an accuracy of 95% when tested in an obstacle avoidance problem on mobile robotics with a minimum use of resources.

Neuromorphic perception with event-based sensors, asynchronous hardware, and spiking neurons shows promise for real-time, energy-efficient inference in embedded systems. Brain-inspired computing aims to enable adaptation to changes at the edge with online learning. However, the parallel and distributed architectures of neuromorphic hardware based on co-localized compute and memory imposes locality constraints to the on-chip learning rules. We propose the event-based three-factor local plasticity (ETLP) rule that uses the pre-synaptic spike trace, the post-synaptic membrane voltage and a third factor in the form of projected labels with no error calculation, that also serve as update triggers. ETLP is applied to visual and auditory event-based pattern recognition using feedforward and recurrent spiking neural networks. Compared to back-propagation through time, eProp and DECOLLE, ETLP achieves competitive accuracy with lower computational complexity. We also show that when using local plasticity, threshold adaptation in spiking neurons and a recurrent topology are necessary to learn spatio-temporal patterns with a rich temporal structure. Finally, we provide a proof of concept hardware implementation of ETLP on FPGA to highlight the simplicity of its computational primitives and how they can be mapped into neuromorphic hardware for online learning with real-time interaction and low energy consumption.

Purpose: Fillet welding is one of the most widespread types of welding in the industry, which is still carried out manually or automated by contact. This paper aims to describe an online programming system for noncontact fillet welding robots with “U”- and “L”-shaped structures, which responds to the needs of the Fourth Industrial Revolution. Design/methodology/approach: In this paper, the authors propose an online robot programming methodology that eliminates unnecessary steps traditionally performed in robotic welding, so that the operator only performs three steps to complete the welding task. First, choose the piece to weld. Then, enter the welding parameters. Finally, it sends the automatically generated program to the robot. Findings: The system finally managed to perform the fillet welding task with the proposed method in a more efficient preparation time than the compared methods. For this, a reduced number of components was used compared to other systems: a structured light 3 D camera, two computers and a concentrator, in addition to the six-axis industrial robotic arm. The operating complexity of the system has been reduced as much as possible. Practical implications: To the best of the authors’ knowledge, there is no scientific or commercial evidence of an online robot programming system capable of performing a fillet welding process, simplifying the process so that it is completely transparent for the operator and framed in the Industry 4.0 paradigm. Its commercial potential lies mainly in its simple and low-cost implementation in a flexible system capable of adapting to any industrial fillet welding job and to any support that can accommodate it. Originality/value: In this study, a robotic robust system is achieved, aligned to Industry 4.0, with a friendly, intuitive and simple interface for an operator who does not need to have knowledge of industrial robotics, allowing him to perform a fillet welding saving time and increasing productivity.

Resumen La visión artificial está cobrando cada día más auge en el mundo de la robótica industrial, ya que es necesario realizar tareas cada vez más precisas y autónomas, por lo que se necesita un posicionamiento del robot más exacto. Para ello se precisa del apoyo de un sistema de visión que sea el que preste al robot precisión en su pose, calibrando dicho sistema con respecto al robot. Este trabajo presenta una metodología sencilla para abordar esta forma de calibración, llamada ojo a mano, empleando una cámara 3D de luz estructurada que obtiene la información del mundo real y un brazo robótico industrial de seis ejes. Esto permite utilizar el algoritmo RANSAC para la determinación de los planos, cuya intersección nos da las coordenadas de los puntos,lo que supone una reducción notable de los errores, ya que las coordenadas proceden de planos ajustados a miles de puntos, lo cual hace que el sistema sea más robusto y capaz de obtener una matriz de transformación de las coordenadas de la cámara a la base del robot, que le permitirá abordar cualquier tarea que precise con una precisión eficiente. Se ha realizado el análisis de errores resultante utilizando dos cámaras 3D diferentes: una básica (Kinect 360) y otra industrial (Zivid ONE+ M).