Science AI Journal
Computer Science · Research topic

Open research questions in Context-Aware Activity Recognition Systems

54 unresolved questions extracted from the limitations and future-work sections of 228 Context-Aware Activity Recognition Systems papers in our library. Each links back to the study that raised it.

What the literature leaves open

  • Existing monitoring solutions have limitations such as user non-compliance and privacy invasion. There is a need for a reliable and efficient monitoring system for elderly individuals.

    Safe Sound: AI-Based Elderly Sound Safety Monitoring System · 2026 · DOI

  • Latency and connectivity constraints in remote or forested areas. High power consumption of traditional systems. Limited ability of conventional systems to capture animal behavioral and emotional states.

    Intelligent NB-IoT and Emotion-Aware Monitoring System for Proactive Wildlife Security · 2026 · DOI

  • Multimodal artifacts, when artifacts of one sensor affect feature representations of other sensors - Asynchronous failure and partial observability, which can cause instability in multimodal representations - Limited energy, heat, computational capacity, and physiological acceptability in wearable body area networks

    Multimodal Sensing and Artificial Intelligence–Driven Data Fusion in Wearable Health Technologies, Advances, System Challenges, and Research Frontiers · 2026 · DOI

  • Limited ecological validation. Small sample sizes. The persistent simulation gap. Technical challenges related to data synchronization, sensor calibration, and algorithmic interpretability.

    Predicting Falls in Older Adults using AI-Powered Wearable IoT Devices: A Scoping Review · 2026 · DOI

  • The complexity of human movement complicates accurate capture and interpretation. Prior work has provided meaningful advancements, but several issues still remain regarding the degree of adaptability and accuracy of feedback.

    Movement Recognition Algorithm Empowered Wearable Devices on Internet of Things Platform for Sports Training · 2026 · DOI

  • The independent effects of each component cannot be separately isolated in the experimental design. The study was conducted in a specific university setting with a limited number of participants.

    Construction and Quantitative Evaluation of Basketball Smart Teaching Scenarios Integrating Ideological and Political Elements · 2026 · DOI

  • The integration effectiveness and scientific evaluation mechanism of ideological and political elements with sports skills have not yet been effectively established. Current teaching practices lack a data-driven technical framework with a closed-loop feedback mechanism.

    Construction and Quantitative Evaluation of Basketball Smart Teaching Scenarios Integrating Ideological and Political Elements · 2026 · DOI

  • Current AI-enabled physical education systems rely on static rules or fixed model architectures, making it difficult to dynamically capture student characteristics. There is a need for a unified methodology for multimodal signal acquisition, adaptive information processing, and closed-loop feedback optimization.

    AI-Driven Personalized Physical Education and Improvement of Key Physical Fitness Characteristics · 2026 · DOI

  • Limited wearability and insufficient motion feature extraction accuracy of traditional dance motion capture systems. Lack of real-time interaction mechanisms in traditional dance instruction.

    Design of a Real-Time Interactive System for Dance Motion Feature Extraction Based on Flexible Textile Sensor Networks · 2026 · DOI

  • The paper does not provide a comprehensive evaluation of the proposed framework. It does not address the potential privacy risks of using wearable textile interfaces.

    Research Review on Age-Friendly Interaction Design in Smart Home Environments: From Home Textile Interfaces to Mobile Feedback · 2026 · DOI

  • Further research is needed to address the challenges of signal noise suppression and body motion artifact separation in bedding textile interfaces. The development of individualized calibration mechanisms for sofa and seating interfaces is necessary. Research on washing durability and flexible connection reliability in wearable textile interfaces is required.

    Research Review on Age-Friendly Interaction Design in Smart Home Environments: From Home Textile Interfaces to Mobile Feedback · 2026 · DOI

  • Kinematic-to-physical ambiguity. Limited supervision for distinguishing physically distinct outcomes. Residual dataset bias remains a limiting factor under domain shift.

    Bridging the visual-to-physical gap: physically aligned representations for fall risk analysis · 2026 · DOI

  • Residual dataset bias remains a limiting factor under domain shift. The effect size of PHARL is domain-dependent, with some datasets showing only modest changes.

    Bridging the visual-to-physical gap: physically aligned representations for fall risk analysis · 2026 · DOI

  • The proposed system can be enhanced with machine learning algorithms to improve detection accuracy. The system can be integrated with health monitoring sensors to track user's vital parameters.

    Design of a Wearable Embedded Device for Real-Time Fall Detection and Emergency Alert for the Elderly · 2026 · DOI

  • There is a need for a wearable and real-time fall detection system. Traditional fall detection methods have limitations such as privacy risks and high costs.

    Design of a Wearable Embedded Device for Real-Time Fall Detection and Emergency Alert for the Elderly · 2026 · DOI

  • Existing CSI-based HAR approaches suffer from high signal volatility, limited feature representation capacity, and poor adaptability in occluded or dynamic environments. There is a need for a robust and scalable solution for next-generation human activity recognition.

    Edge-Enabled Human Activity Recognition Using Hybrid Deep Learning and Multisensor Wi-Fi CSI Arrays · 2026 · DOI

  • HAR plays a vital role in health monitoring, smart healthcare, and related fields. While vision-based methods are hindered by high deployment costs and privacy concerns, and wearable sensor solutions suffer from contact-based limitations and discomfort, Wi-Fi CSI-based approaches offer a promising nonintrusive alternative. However, existing solutions often rely on cloud-based processing and struggle in occluded or NLOS scenarios, limiting their real-world deployment. To address these limitations, we developed a low-cost microcontroller. The system synchronously collects multichannel CSI data, significantly improving signal stability and Wi-Fi CSI array acquisition system based on the ESP32 2 × 2 reducing loss, as evidenced by improvements in statistical volatility, MAE, and ADF test outcomes compared to single- receiver baselines. By applying both FFT and wavelet transforms, we captured complementary time–frequency features, enhancing the model’s capacity to recognize complex motion patterns. Furthermore, we proposed a hybrid DL framework deployed on the STM32H7 MCU to enable on-device edge inference. Compared with conventional models (e.g., CNN, LSTM, BP, and SVM), the C–L–A model demonstrated superior generalization, attention-driven feature refinement, and robustness, achieving up to 99% accuracy under LOS and 97% under NLOS conditions. Overall, this study presents a complete edge-deployable Wi-Fi CSI-based HAR system, spanning data acquisition, feature extraction, and real-time recognition. Future work will focus on three key directions to further support practical applications in smart homes, elderly care, and rehabilitation scenarios: first, lightweight model optimization to reduce computational overhead; second, adaptive feature selection to enhance recognition robustness; and third, the integration of internet of things (IoT) transmission security mechanisms alongside energy-efficient deployment strategies— especially critical for ensuring reliable operation under low-latency and resource-constrained IoT conditions [41], [42]. ACKNOWLEDGMENT The authors would like to express their sincere gratitude to the editor and the anonymous reviewers for their insightful and constructive comments, which have significantly improved the quality of this article. The authors also thank all colleagues and participants who contributed to the data collection and experimental validation of this research.

    Edge-Enabled Human Activity Recognition Using Hybrid Deep Learning and Multisensor Wi-Fi CSI Arrays · 2026 · DOI

  • Wearable Sensor-based Elderly Activity Recognition (WSEAR) remains difficult due to the diversity of activities and overlapping or uncertain activities. Existing methods have limitations in recognizing complex and similar activities.

    Deep learning and ontology-fuzzy inference system for elderly activity recognition · 2026 · DOI

  • The current systems do not provide dynamic and supportive health surveillance and fall prevention for the elderly. The current systems do not enable remote monitoring through a mobile application.

    An Integrated AI And IoT Based System for Elderly Fall Detection and Health Monitoring · 2026 · DOI

  • Vision Transformers for improved occlusion and pose robustness via knowledge distillation-based compression. Federated learning across distributed HSM deployments enabling shared model improvements without centralizing biometric data. Contact-free photoplethysmography for physiological sensing enabling affective computing applications.

    Hybrid Smart Mirror · 2026 · DOI

  • Existing smart mirror implementations are constrained by single-modality designs, absent user identification, insecure IoT integration, and inadequate empirical evaluation. There is a need for a multi-modal ambient intelligence platform that fuses real-time biometric identification, natural language voice interaction, and IoT device orchestration.

    Hybrid Smart Mirror · 2026 · DOI

  • Conventional wildlife protection systems lack the ability to capture animal behavioral and emotional states. Traditional techniques lack the capacity to recognize animal conduct in real-time. There is a need for a system that integrates emotion recognition and adaptive geofencing to enhance wildlife security.

    Intelligent NB-IoT and Emotion-Aware Monitoring System for Proactive Wildlife Security · 2026 · DOI

  • Traditional early warning systems often fail to identify behavioral markers predictive of declining involvement. Conventional techniques of evaluating academic progress fail to capture the nuances of students' behavior and emotional states during instruction.

    BEACON-AI: A CNN- BiLSTM -Attention Framework for Real-Time Multimodal Student Behavior Analysis and Academic Early Warning · 2026 · DOI

  • There is a need for a system that can detect tremor levels in patients with Parkinson's disease with high accuracy. Previous studies have limitations in terms of accuracy and reliability.

    Machine Learning-Based Real-Time Tremor Level Detection for Parkinson Disease · 2026 · DOI

  • The study identifies misclassification patterns, particularly for rail-based transportation modes and vehicles at traffic lights or stops. The exclusion of high-energy sensors like GPS can lead to reduced accuracy. The study highlights the need for careful consideration of model training and evaluation due to imbalanced datasets.

    Power-Aware Transport Mode Detection: A Comparative Analysis on Resource-Constrained Smartphones · 2026 · DOI

Most-cited papers in Context-Aware Activity Recognition Systems

Most recent work

Find a gap in your own Context-Aware Activity Recognition Systems sub-topic

This page shows what the Context-Aware Activity Recognition Systems literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.

Open the Research Gap Finder →

Related topics in Computer Science

Online Learning and Analytics
851 open questions
Digital literacy in education
518 open questions
AI in Service Interactions
517 open questions
Music Technology and Sound Studies
423 open questions
Teaching and Learning Programming
384 open questions
Quantum Computing Algorithms and Architecture
368 open questions
Virtual Reality Applications and Impacts
317 open questions
Intelligent Tutoring Systems and Adaptive Learning
298 open questions
54 open questions have been extracted from the limitations and future-work passages of 228 Context-Aware Activity Recognition Systems papers in our library. Each one below links back to the study that raised it, so you can read the original claim in context.

Tools for your next paper

Gaps & Literature SearchFind open questions in any field — 120,000+ open gaps mined from our own library.Pre-CheckIs your paper ready? A Tier 1–5 acceptance probability in seconds.Journal FinderA ranked shortlist of target journals from a 17,500-venue index.Duplicate Publication CheckerCheck prior-publication & salami-slicing risk across 8 sources.Citation GeneratorFreeA DOI or title → APA, MLA, Vancouver, Chicago, BibTeX, RIS. No signup.Graphical Abstract MakerFreeTurn your findings into a colorblind-safe graphical abstract.

Compare the category — Honest roundups of the AI research tools, ours listed alongside the alternatives.

Command palette

Jump anywhere, run any action.