Cognitive Fatigue (CF) is the decline in cognitive abilities due to prolonged exposure to mentally demanding tasks. In this paper, we used gait cycle analysis, a biometric method related to human locomotion to identify cognitive fatigue in individuals. The proposed system in this paper takes two asynchronous videos of the gait of individuals to classify if they are cognitively fatigued or not. We leverage the pose estimation library OpenPose, to extract the body keypoints from the frames in the videos. To capture the spatial and temporal information of the gait cycle, a CNN-based model is used in the system to extract the embedded features which are then used to classify the cognitive fatigue level of individuals. To train and test the model, a gait dataset is built from 21 participants by collecting walking data before and after inducing cognitive fatigue using clinically used games. The proposed model can classify cognitive fatigue from the gait data of an individual with an accuracy of 81%.

We present a method for extracting high-level semantic information through successful landmark detection using feature fusion between RGB and depth information. We focus on the classification of specific labels (open path, humans, staircases, doorways, obstacles) in the encountered scene, which can be a fundamental source of information enhancing scene understanding, and acting towards the safe navigation of the mobile unit. Experiments are conducted using a manual wheelchair equipped with a stereo RGB-D camera that captures image instances consisting of multiple labels before fine-tuning on a pre-trained Vision Transformer (ViT).

This paper describes a novel shared control teleoperation framework for mobile robots that utilizes Control Barrier Functions (CBFs) as filtering mechanism to prevent a human operator from making dangerous actions. The proposed framework demonstrates the potential to create a CBF controller that enables users with no prior knowledge of robotics to safely tele-navigate mobile robots with limited situational awareness. As formal methods, we utilize a hand-crafted CBF, which acts as a repulsive field to describe unsafe regions withing the robot’s vicinity. The implementation of the application was deemed possible by creating a Virtual Reality (VR) simulation in the Unity Engine with the SUMMIT-XL STEEL mobile base as an experimental platform. Preliminary experimental results show the ability of the framework to enable safe teleoperation.

We introduce an innovative assistive robotic setup tailored for assessing the cognitive state of individuals with spinal cord injuries (SCI) during their daily activities. Utilizing physiological sensors such as ECG, EEG, and EDA, along with cameras for facial expression, our system is designed to detect and evaluate cognitive fatigue in participants as they engage with a collaborative robot. Specifically, two tasks, Cooking Pasta Sauce and Getting Ready for Work–have been crafted to gather data on cognitive states. Participants interact with the robot using natural language (English) to perform tasks, while their physiological responses, facial expressions, and activities are recorded. The study comprises three phases of cognitive fatigue: baseline, moderate, and severe. Cognitive fatigue is induced through the N-back task paradigm, and its severity is assessed using the Visual Analogue Scale for Fatigue (VAS-F) questionnaire. Our system is designed to intervenes during the tasks based on the detected cognitive fatigue levels. In this paper, we concentrate on validating the cognitive fatigue detection system using only physiological sensors during task performance, achieving an accuracy of 85.7% and a recall of 0.87. We provide detailed insights into the system design and present a preliminary analysis of the gathered data.

This article presents a method for extracting high-level semantic information through successful landmark detection using 2D RGB images. In particular, the focus is placed on the presence of particular labels (open path, humans, staircase, doorways, obstacles) in the encountered scene, which can be a fundamental source of information enhancing scene understanding and paving the path towards the safe navigation of the mobile unit. Experiments are conducted using a manual wheelchair to gather image instances from four indoor academic environments consisting of multiple labels. Afterwards, the fine-tuning of a pretrained vision transformer (ViT) is conducted, and the performance is evaluated through an ablation study versus well-established state-of-the-art deep architectures for image classification such as ResNet. Results show that the fine-tuned ViT outperforms all other deep convolutional architectures while achieving satisfactory levels of generalization.