With the widespread adoption of deep learning, reinforcement learning (RL) has experienced a dramatic increase in popularity, scaling to previously intractable problems, such as playing complex games from pixel observations, sustaining conversations with humans, and controlling robotic agents. However, there is still wide range domains inaccessible RL due high cost danger interacting environment. Offline paradigm that learns exclusively static datasets collected interactions, making it...

This paper proposes two new data augmentation approaches based on Deep Convolutional Generative Adversarial Networks (DCGANs) and Style Transfer for augmenting Parkinson’s Disease (PD) electromyography (EMG) signals. The experimental results indicate that the proposed models can adapt to different frequencies amplitudes of tremor, simulating each patient’s tremor patterns extending them sets movement protocols. Therefore, one could use these existing patient dataset generating simulations...

Parkinson's disease (PD) is a neurological disorder requiring early and accurate diagnosis for effective management. Machine learning (ML) has emerged as powerful tool to enhance PD classification diagnostic accuracy, particularly by leveraging wearable sensor data. This survey comprehensively reviews current ML methodologies used in classifying Parkinsonian tremors, evaluating various tremor data acquisition methodologies, signal preprocessing techniques, feature selection methods across...

The ability to communicate with robots using natural language is a significant step forward in human-robot interaction. However, accurately translating verbal commands into physical actions promising, but still presents challenges. Current approaches require large datasets train the models and are limited maximum of 6 degrees freedom. To address these issues, we propose framework called InstructRobot that maps instructions robot motion without requiring construction or prior knowledge...

Autonomous artificial agents must be able to learn behaviors in complex environments without humans design tasks and rewards. Designing these functions for each environment is not feasible, thus, motivating the development of intrinsic reward functions. In this paper, we propose using several cognitive elements that have been neglected a long time build an internal world model intrinsically motivated agent. Our agent performs satisfactory iterations with environment, learning needing...

Aerial platforms, such as quadrotors, are inherently unstable systems. Generally, the task of stabilizing flight a quadrotor is approached by techniques based on classic and modern control algorithms. However, recent model-free reinforcement learning algorithms have been successfully used for controlling drones. In this work we show feasibility applying methods to optimize stochastic policy (during training), in order perform position "model-free" quadrotor. This process achieved while...

Partial hand amputations are common in developing countries and have a negative impact on patients their families' quality of life. The uniqueness each partial amputation, coupled with the relatively high costs prostheses, makes it challenging to provide suitable prosthetic solutions countries. Current often long lead times require level expertise produce. aim this study was design develop an affordable patient-specific prosthesis for countries.The designed patient transmetacarpal amputation...

Abstract Recent advancements in AI and deep learning have created a growing demand for artificial agents capable of performing tasks within increasingly complex environments. To address the challenges associated with continuous constraints knowledge capacity this context, cognitive architectures inspired by human cognition gained significance. This study contributes to existing research introducing cognitive-attentional system employing constructive neural network-based approach acquisition...

Understanding consciousness is one of the most fascinating challenges our time. From ancient civilizations to modern philosophers, questions have been asked on how conscious his/her own existence and about world that surrounds him/her. Although there no precise definition for consciousness, an agreement it strongly related human cognitive processes such as attention, a process capable promoting selection few stimuli from huge amount information reaches us constantly. In order bring...

With the widespread adoption of deep learning, reinforcement learning (RL) has experienced a dramatic increase in popularity, scaling to previously intractable problems, such as playing complex games from pixel observations, sustaining conversations with humans, and controlling robotic agents. However, there is still wide range domains inaccessible RL due high cost danger interacting environment. Offline paradigm that learns exclusively static datasets collected interactions, making it...

This article introduces the successful methodology adopted for realization of Brazilian Robotics Olympiad. Organizational and scientific issues are presented discussed as well results statistical surveys performed with participants that verify benefits brought by this robotics The Olympiad is a nine-year-old nation-wide initiative created mission promoting among students or without previous knowledge robotics, fostering their interest to engage in science, technology engineering studies...

This paper proposes a comparison between different neural network models, using multilayer perceptron (MLPs) and recurrent (RNN) for predicting Parkinson's disease electromyography (EMG) signals, to anticipate resulting resting tremor patterns. The experimental results indicate that the proposed models can adapt frequencies amplitudes of tremor, provide reasonable predictions both EMG envelopes raw signals. Therefore, one could use these as input control strategy functional electrical...

Controlling a humanoid robot with its typical many degrees of freedom is complex task, and methods have been proposed to solve the problem locomotion. In this work, we generate gait for Hitec Robonova-I using model-free approach, where fairly simple parameterized models, based on truncated Fourier series, are applied joint angular trajectories. To find parameter set that generates fast stable walk, optimization algorithms were used, specifically genetic algorithm particle swarm optimization....

The increase of applications that use autonomous robots has endowed them with a high number sensors and actuators are sometimes redundant. This new highly complex systems the type environment where they expected to operate require deal data overload fusion. In humans face same problem when sounds, images, smells presented their in daily scene, natural filter is applied: attention. Although there many computational models apply attentive robotics, usually restricted two classes systems: 1)...

This paper shows that CIDEr-D, a traditional evaluation metric for image description, does not work properly on datasets where the number of words in sentence is significantly greater than those MS COCO Captions dataset. We also show CIDEr-D has performance hampered by lack multiple reference sentences and high variance length. To bypass this problem, we introduce CIDEr-R, which improves making it more flexible dealing with length variance. demonstrate CIDEr-R accurate closer to human...

Data derived from the realm of social sciences is often produced in digital text form, which motivates its use as a source for natural language processing methods. Researchers and practitioners have developed relied on artificial intelligence techniques to collect, process, analyze documents legal field, especially tasks such summarization classification. While increasing procedural efficiency primary motivation behind several works proposed solutions human rights-related issues, assessment...

Automatically describing images using natural sentences is essential to visually impaired people’s inclusion on the Internet. This problem known as Image Captioning. There are many datasets in literature, but most contain only English captions, whereas with captions described other languages scarce. We introduce #PraCegoVer, a multi-modal dataset Portuguese based posts from Instagram. It first large for image captioning Portuguese. In contrast popular datasets, #PraCegoVer has one reference...

Unmanned Aerial Vehicles (UAVs), or drones, have recently been used in several civil application domains from organ delivery to remote locations wireless network coverage. These platforms, however, are naturally unstable systems for which many different control approaches proposed. Generally based on classic and modern control, these algorithms require knowledge of the robot's dynamics. However, recently, model-free reinforcement learning has successfully controlling drones without any prior...