Human Action Research Articles

Real-Time EtherCAT-Based Control Architecture for Electro-Hydraulic Humanoid

Electro-hydraulic actuators have witnessed significant development over recent years due to their remarkable abilities to perform complex and dynamic movements. Integrating such an actuator in humanoids is highly beneficial, leading to a humanoid capable of performing complex tasks requiring high force. This highlights the importance of safety, especially since high power output and safe interaction seem to be contradictory; the greater the robot’s ability to generate high dynamic movements, the more difficult it is to achieve safety, as this requires managing a large amount of motor energy before, during, and after the collision. No matter what technology or algorithm is used to achieve safety, none can be implemented without a stable control system. Hence, one of the main parameters remains the quality and reliability of the robot’s control architecture through handling a huge amount of data without system failure. This paper addresses the development of a stable control architecture that ensures, in later stages, that the safety algorithm is implemented correctly. The optimum control architecture to utilize and ensure the maximum benefit of electro-hydraulic actuators in humanoid robots is one of the important subjects in this field. For a stable and safe functioning of the humanoid, the development of the control architecture and the communication between the different components should adhere to some requirements such as stability, robustness, speed, and reduced complexity, ensuring the easy addition of numerous components. This paper presents the developed control architecture for an underdeveloped electro-hydraulic actuated humanoid. The proposed solution has the advantage of being a distributed, real-time, open-source, modular, and adaptable control architecture, enabling simple integration of numerous sensors and actuators to emulate human actions and safely interact with them. The contribution of this paper is an enhancement of the updated rate compared to other humanoids by 20% and by 40 % in the latency of the master. The results demonstrate the potential of using EtherCAT fieldbus and open-source software to develop a stable robot control architecture capable of integrating safety and security algorithms in later stages.

A capabilitarian behavioral economics: what behavioral economics can learn from the capability approach

AbstractBehavioral economics (BE) accounts for how people actually behave has proven to be very influential in multiple domains. Its insights have dislodged settled assumptions and put forward an improved understanding of human beings and their agency. Although it challenges the conventional approach, advanced by rational choice theory (RCT), because it also formally builds on it, BE shares some of RCT’s advantages as well as its shortcomings. One the one hand, it shares an important focus on freedom. On the other, it is limited to self-interest as the main motivation for human action. The latter is a limitation since human experience is much richer than that. To complement BE’s limitation, this paper suggests Amartya Sen’s Capability Approach (CA). This framework places humans and their quality of life at the locus of attention for the assessment of social states. To do so, it challenges RCT in a more fundamental way and it provides a richer account of human agency, which includes both self-regarding as well as other-regarding motivations. This combination enriches BE’s realism as well as practicality.

Is it personal or is it social? The interaction of knowledge domain and statistical evidence in U.S. and Chinese preschoolers' social generalizations.

Children make inferences about the social world by observing human actions. However, human actions can be ambiguous: They can be sources of information about personal, idiosyncratic characteristics of individuals or socially shared knowledge. In two cross-cultural studies (N = 420; Mage = 4.05 years, SD = 0.77, 47% female), we ask if U.S. and Chinese children's inferences about whether an action is personal or social vary by domain, statistical evidence, and culture. We did this with a generalization method: Preschoolers learn about one agent's actions and then are asked what they think a new agent will do. Low rates of generalization suggest children inferred something unique to an individual, while high rates suggest that children inferred that the action represented socially shared knowledge. In a mixed between- and within-participant design, children observed agents demonstrate sequences of statistically random (or nonrandom, between participants) actions that were verbally framed as relevant to a particular domain (agent's personal preferences, labels, object functions, or game rules). We found that children's social generalizations about actions were on a continuum: with linguistic conventions (e.g., labels) being the most social, preferences being the most personal, and nonlinguistic conventions (i.e., object functions, game rules) falling somewhere in between. Furthermore, the influence of statistical evidence and cultural variation varied for each domain. These findings highlight how children combine knowledge and evidence to infer social meaning from actions and have implications for rational constructivist accounts of cultural learning. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

VT-Grapher: Video Tube Graph Network With Self-Distillation for Human Action Recognition

VT-Grapher: Video Tube Graph Network With Self-Distillation for Human Action Recognition

The Deep Learning-based Human Action Recognition System for Competitive Sports

The Deep Learning-based Human Action Recognition System for Competitive Sports

Action-Aware Vision Language Navigation (AAVLN): AI vision system based on cross-modal transformer for understanding and navigating dynamic environments

Visually impaired individuals face great challenges with independently navigating dynamic environments because of their inability to fully comprehend the environment and actions of surrounding people. Conventional navigation approaches like Simultaneous Localization And Mapping (SLAM) rely on complete scanned maps to navigate static, fixed environments. With Vision Language Navigation (VLN), agents can understand semantic information to expand navigation to similar environments. However, both cannot accurately navigate dynamic environments containing human actions. To address this challenge, we propose a novel cross-modal transformer-based Action-Aware VLN system (AAVLN). Our AAVLN Agent Algorithm is trained using Reinforcement Learning in our Environment Simulator. AAVLNs novel cross-modal transformer structure allows the Agent Algorithm to understand natural language instructions and semantic information for navigating dynamic environments and recognizing human actions. For training, we use Reinforcement Learning in our action-based environment simulator. We created it by combining an existing simulator with our novel 3D human action generator. Our experimental results demonstrate the effectiveness of our approach, outperforming current methods on various metrics across challenging benchmarks. Our ablation studies also highlight that we increase dynamic navigation accuracy with our Vision Transformer based human action recognition module and cross-modal encoding. We are currently constructing 3D models of real-world environments, including hospitals and schools, for further training AAVLN. Our project will be combined with Chat-GPT to improve natural language interactions. AAVLN will have numerous applications in robotics, AR, and other computer vision fields.

Exploring High-Order Skeleton Correlations with Physical and Non-Physical Connection for Action Recognition

Hypergraphs have received widespread attention in modeling complex data correlations due to their superior performance. In recent years, some researchers have used hypergraph structures to characterize complex non-pairwise joints in the human skeleton and model higher-order correlations of the human skeleton. However, traditional methods of constructing hypergraphs based on physical connections ignore the dependencies among non-physically connected joints or bones, and it is difficult to model the correlation among joints or bones that are highly correlated in human action but are physically connected at long distances. To address these issues, we propose a skeleton-based action recognition method for hypergraph learning based on skeleton correlation, which explores the effects of physically and non-physically connected skeleton information on accurate action recognition. Specifically, in this paper, spatio-temporal correlation modeling is performed on the natural connections inherent in humans (physical connections) and the joints or bones that are more dependent but not directly connected (non-physical connection) during human actions. In order to better learn the hypergraph structure, we construct a spatio-temporal hypergraph neural network to extract the higher-order correlations of the human skeleton. In addition, we use an attentional mechanism to compute the attentional weights among different hypergraph features, and adaptively fuse the rich feature information in different hypergraphs. Extensive experiments are conducted on two datasets, NTU-RGB+D 60 and Kinetics-Skeleton, and the results show that compared with the state-of-the-art skeleton-based methods, our proposed method can achieve an optimal level of performance with significant advantages, providing a more accurate environmental perception and action analysis for the development of embodied intelligence.

Shifting fire regimes cause continent-wide transformation of threatened species habitat.

Human actions are causing widespread increases in fire size, frequency, and severity in diverse ecosystems globally. This alteration of fire regimes is considered a threat to numerous animal species, but empirical evidence of how fire regimes are shifting within both threatened species' ranges and protected areas is scarce, particularly at large spatial and temporal scales. We used a big data approach to quantify multidecadal changes in fire regimes in southern Australia from 1980 to 2021, spanning 415 reserves (21.5 million ha) and 129 threatened species' ranges including birds, mammals, reptiles, invertebrates, and frogs. Most reserves and threatened species' ranges within the region have experienced declines in unburnt vegetation (≥30 y without fire), increases in recently burnt vegetation (≤5 y since fire), and increases in fire frequency. The mean percentage of unburnt vegetation within reserves declined from 61 to 36% (1980 to 2021), whereas the mean percentage of recently burnt vegetation increased from 20 to 35%, and mean fire frequency increased by 32%, with the latter two trends primarily driven by the record-breaking 2019 to 2020 fire season. The strongest changes occurred for high-elevation threatened species, and reserves of high elevation, high productivity, and strong rainfall decline, particularly in the southeast of the continent. Our results provide evidence for the widely held but poorly tested assumption that threatened species are experiencing widespread declines in unburnt habitat and increases in fire frequency. This underscores the imperative for developing management strategies that conserve fire-threatened species in an increasingly fiery future.

Cooked or discarded? Experimental distinction of rabbit burnt bones and its application to the archaeological record

Leporid bones from Middle Palaeolithic assemblages may result from the contributions of various predators, including humans, or natural processes. Although the acquisition, processing, and consumption of small mammals by Neanderthal groups has been widely demonstrated, there are still some unresolved issues. Cut marks are typically the most reliable indicators of human processing of carcasses. However, animals the size of rabbits often pose a challenge in this regard, as the use of stone tools is not always necessary for their consumption, particularly for meat or marrow extraction. Consequently, the quantity of these types of marks, such as cuts or hammer-stone percussion damage, indicating human processing, may be limited. Burning is another type of bone damage that could be indicative of human action, although both intentional and accidental processes could lead to the thermal alteration of remains without necessarily linking them to consumption. Therefore, efforts to distinguish the processes resulting in bone burning are of vital importance in determining the origin of these animals in archaeological assemblages. In this work, the results of several experimental series designed to characterise the roasting and subsequent cleaning of waste on rabbit bones are presented. These results confirm most of the characteristics described in previous experimental works on burnt bones, highlighting the differential damage between bones with and without meat. The current study aims to contribute new data for characterising burned rabbit bones resulting from human actions, which can then be applied to Middle Palaeolithic assemblages with this type of thermal alteration bone modifications.

Discerning Deception: A Face-Centric Deepfake Detection Approach with ResNeXt-50 and LSTMs

Abstract: A.I. has grown to epidemic proportions over the last years as its applied in almost all sectors to allocate workload from humans but end up being done effectively with no human intervention. A branch of A.I. called deep learning, which operates by mimicking human judgment and action through neural network systems. Nonetheless, with the increase height of the two platforms have been experienced sufficient cases of misguided individuals using tools to recycle videos, audios, and texts to achieve their agendas. This insinuates a due assumption that Generative Adversarial Networks, GANs, are central to the development of believable deepfakes. GANs have developed a crucial ability to generate videos that replace frames with material from another video source to create deepfakes videos. While GANs serves various purposes such as entertainment, teaching, and experimentation, malicious actors can misuse these deep learning techniques to manipulate videos, impacting the privacy of individuals in society. This paper conducts an analysis of different deepfake detection models, comparing their efficacy and discussing potential future extensions of deepfake technology. The study presents a novel deepfake detection approach utilizing a combination of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN). This method utilizes ResNext50 for extracting features at the frame level, while employing LSTM (Long Short-Term Memory) for video classification based on these extracted features. Various datasets are incorporated, including the deepfake detection challenge dataset (DFDC) and Face Forensics deepfake collections (FF++), combining them to achieve a high-accuracy model capable of accurately discerning between real and deepfake videos. The results of this study make a valuable contribution to the continuous endeavors aimed at improving deepfake detection abilities and ensuring privacy protection in a time heavily influenced by artificial intelligence

Enable Flexibilisation in FAIRWork’s Democratic AI-based Decision Support System by Applying Conceptual Models Using ADOxx

Decision-making in complex production environments is challenging as the information and knowledge requirements must be constantly observed since the ecosystems they operate in are continuously changing. Artificial intelligence (AI) can tackle complexity in decision-making by making machines more intelligent. But reacting to changing or new problems and related decision processes to facilitate the understanding of the involved humans is an equally important problem. Therefore, decision support systems are required to assist complex decisions and enable flexibility to support the decision-makers. Within this scope, we will introduce the Democratic AI-based Decision Support System (DAI-DSS), which is designed and implemented within the EU-funded FAIRWork project, considering both human and machine actors during decision-making. The FAIRWork project proposes a model-based approach to both express high-level decision scenarios and formally describe the decision processes, which are then used as input for configuring the decision support system to meet concrete decision problems.

Ethical Issues on Cloning and Genetically Modified Products

Science and Technology have greatly improved man’s quality of life as new discoveries are designed to nurture a healthy and long life. Science and Technology can be described as two sides of a coin as science discovers the problem and technology provides the product to solve that particular problem that affects the well-being of mankind. Philosophy, like science, is a discipline that is hinged on rational thinking and the pursuit of truth. The branch of philosophy that is concerned with how man ought to live in the society is Ethics. Ethics is the branch of philosophy that deals with rightness and wrongness of human actions. It also determines the rightness or wrongness of technological innovations and their effect on the health of man. There are numerous scientific discoveries in the area of genetic engineering but this paper is limited to cloning and genetic modification. In this work, ethical implications of these scientific discoveries were identified.  This paper shall use the method of conceptual analysis. It intends to address the ethical implications of cloning and genetically modified products for mankind. The paper concludes that the benefits of these scientific developments should be embraced as they outweigh the cons and provide succour to beneficiaries which is the objective of Science and Technology. The welfare of mankind is the concern of ethics as well as science and technology.

Multi-Modal Multi Feature Assisted Human Action Recognition through Deep Learning

Multi-Modal Multi Feature Assisted Human Action Recognition through Deep Learning

GIS-based relationship between pathway names and landscape. A multilingual case study: Euskadi, Spain

Place names determine geographic units in space, encapsulate the description of places where inhabitants conduct their activities, and preserve the perception of the territory in past times. The very definition of landscape impacts two interrelated concepts: human action and perception. On another note, traditional pathways are structuring elements of the landscape, bearing witness to the dynamic relationship between the territory and its inhabitants by connecting residences and workplaces. This work aims to assess the relationship between the landscape mosaic and the toponyms of pathways and roads gathered in the current cartography of Euskadi (Spain), a territory influenced by the existence of two official languages. Given the spatial component of the data, this analysis is conducted through geographic information systems. Firstly, a corpus of 3072 pathway names selected from current official toponymic databases is compiled, as well as the content of the first edition of the National Topographic Map. Subsequently, the semantic content of the corpus elements is examined, as well as the nature of their referential content concerning the landscape units obtained from the Atlas of Spanish Landscapes. The results show common factors in characterizing landscapes and the etymology of names. Thus, it is noted that traditional agricultural, forestry, livestock, and traditional industry activities shape the toponymy in the most populated landscape units. Meanwhile, references to the orography typify the more mountainous landscapes. Therefore, the geographical study demonstrates the existence of common factors that link landscape and toponymy, validating one as a study source for the other.

The Iceberg in the Room: A Causal Layered Analysis of Self with the Futures Sense of Yearning

This Futures focussed paper shares my creative response to a reflexive ricochet of being researcher, to unexpectedly becoming the researched and back again. I am currently using phenomenologically art-led and heuristic research methods to explore the implications of the Futures Senses in tertiary career education. The field of career theory is holistic and multidisciplinary. It draws from psychological, sociological and cultural fields to contextualise human action across the lifespan. Therefore, I work with people in transition and anticipation. This paper emerged from my immersion in the Futures Sense of Yearning and the moment I recognised the futures sense of Yearning in myself. This unexpected epiphany initiated a reflexive response as spectator of, and traveller on, my own transitional and anticipatory experience which was hovering in the spaces between the inner and outer self. Using a Jungian spontaneous drawing technique I set to give voice to what Jung noted as the ‘… two worlds: the world of the external perception and the world of the perception of the unconscious…’. This process revealed latent embodied knowing that surfaced through dialogue with the artwork. The layered emergent data of the artwork is analysed using a Causal Layered Analysis (CLA). The CLA invites diverse and layered perspectives of the subject of inquiry by examining it from the perspectives of litany, systems, worldviews and metaphor. Inayatullah uses an iceberg analogy, to illustrate the layers of an inquiry that are simultaneously exposed and discrete; known, unknown and ‘re-known’. Thus, a CLA of self is used, to reveal the ‘iceburg in the room’, with intention to explicate the meaning of the personal, as it relates to the collective.

An agent architecture for expressive spatial knowledge and reasoning in land use modeling and simulations

This paper presents a new cognitive agent design approach integrating spatial knowledge representation and reasoning in agent-based modeling dedicated to land use simulations. A deep motivation for our agent-centric contribution is the ever-increasing development of spatially explicit agent simulation platforms. We build on this technological evolution and topology theory to endow the agent with human’s spatial representation and reasoning following a Belief–Desire–Intention architecture. A pilot implementation of the methodology with simulation experiments on a hunting model was carried out in GAMA platform to assess agent performances. Simulations display a consistent trend of animal population dynamics and also confirm a high model sensitivity to the integration of spatial knowledge and reasoning in agent-based models of human actor. These results demonstrate a successful implementation and the importance of spatial dimension in the expressive power and the validity of agent-based models. Future research efforts should therefore emphasize on designing human knowledge representation and incorporating learning abilities to improve models efficiency.

Assembling with VR

In this paper, we consider the claims that constellate around the concepts of immersion, presence and empathy that have been made about virtual reality across many disciplines of study, including psychology, criminology, immersive film and media. These claims are applied to an interdisciplinary, collaborative project: VR Dance; which engaged young people (11-16 years) in hip hop and immersive technology workshops over a six-week period. We discuss the ways in which co-created immersive environments which centre the body offer potential to tune into and re-calibrate our sensitivities and modes of engagement with each other and the environments we are in. We argue that this is not simply as a result of technology’s effects on individuals but constituted in wider assemblages of human and nonhuman actors.  We make the case for virtual reality, not as a tool for ‘becoming other’, but as part of wider assemblages in ongoing transformations, relocations, and calibrations. Funding Acknowledgement VR Dance is an East London Dance programme, funded by the Mayor of London’s Young Londoner’s Fund & the London Borough of Redbridge and delivered in partnership with the University of Bristol, BirdGang Ltd, Maskomi and PlayLa.bZ.

Magnetorheological Fluid-Based Haptic Feedback Damper

Damping involves the various frictional and other obstructive effects that attenuate free vibration. For a long time, people have mainly used it to make various dampers to reduce mechanical vibration and consume kinetic energy. It is widely used in fields such as aerospace, automotive, and consumer electronics. These dampers mainly act on mechanical structures. In recent years, with the rapid development of novel human–machine interaction methods and force/tactile feedback technology, the damper has begun to act on people, such as when a person interacts with a robot and their force is applied to a structure with damping. This type of damper requires variable damping, and the amount of variation is controlled by the magnitude of human action. This study used magnetorheological fluid (MRF) instead of traditional damping fluids, such as silicone oil, sesame oil, and mechanical oil. Magnetorheological fluid is a controllable fluid with magnetorheological effects, and its viscosity (hardness) can be changed by changing the nearby magnetic field. This study took the design of variable damping keyboard buttons with haptic feedback as an example to study the electromagnetic and dynamic models of variable dampers based on magnetorheological fluids. The experimental setup was designed and used to verify the haptic effectiveness of the scheme, which can be applied to the development of other haptic dampers that require variable damping.

The Avant-Garde as Ecopoetics, Experimental Landscapes of Filipinx Diasporic Poetry

ABSTRACT The avant-garde is a socially grounded world-making practice, an aesthetic tradition envisioning collectively produced futures. By examining the modernist verses of José García Villa, this article locates Filipinx avant-garde poetry as an ecopoetic site, where nature, the environment, and human actors produce collective-making and consciousness-building exercises. Villa’s poetry does not only connote the bizarreness of being Filipinx American in the early twentieth century, but also illustrates how invoking queerness conjures new worlds outside our precarious moment. Filipinx avant-garde poetry demonstrates why the ecopoetic should not only seek to protect what is scarce but envision what lies beyond our present.

Motion Vector-Based Self-Attention for Real-Time Human Activity Recognition in Compressed Videos: The MVViT Approach

Herein, a novel methodology is proposed for real-time recognition of human activity in a compressed domain of videos based on motion vectors and self-attention mechanism using vision transformers, and it is termed as motion vectors and vision transformers (MVViT). The videos in MPEG-4 and H.264 compression formats are considered for this study. Any video source without any prior setup could be considered by adopting the proposed method to the corresponding video codecs and camera settings. Existing algorithms for recognition of human action in a compressed video have some limitations in this regard, such as (i) requirement of keyframes at a fixed interval, (ii) usage of P frames only, and (iii) normally support single codec only. These limitations are overcome in the proposed method by using arbitrary keyframe intervals, using both P and B frames, and supporting MPEG-4 as well as H.264 codecs. The experimentation is carried out using the benchmark datasets, namely, UCF101, HMDB51, and THUMOS14, and the recognition accuracy in a compressed domain is found to be comparable to that observed in raw video data but at reduced cost of computation. The proposed MVViT method has outperformed other recent methods in terms of a lesser (61.0%) number of parameters and (63.7%) Giga Floating Point Operations Per Second (GFLOPS), while significantly improving accuracy by 0.8%, 5.9% and 16.6% for UCF101, HMDB51 and THUMOS14, respectively. Also, it is observed that the speed is increased by 8% in case of UCF101 when compared to the highest speed reported in the literature on the same dataset. The ablation study of the proposed method has been done using MVViT variants for different codecs and the performance analysis is done in comparison with the state-of-the-art network models.