A Historical View of Active Assisted Living

Sara Colantonio, Slaviša Aleksić, Jean Calleja‐Agius, Kenneth P. Camilleri, Anto Čartolovni, Pau Climent-Pérez, Stefania Cristina, Vladimir Despotović, Hazım Kemal Ekenel, Mustafa Ekrem Erakın, Francisco Flórez-Revuelta, Danila Germanese, Nicole Grech, Murat Emirzeoğlu, Ivo Iliev, Anna Sigríður Íslind, Mladjan Jovanovic, Martin Kampel, William D. Kearns, Andrzej Klimczuk, Lambros Lambrinos, Jennifer Lumetzberger, Wiktor Mucha, Sophie Noiret, Zada Pajalić, Rodrigo Pérez‐Rodríguez, Galidiya Petrova, Sintija Petrovica-Klavina, Peter Počta, Angelica Poli, Mara Romanovska, Albert Ali Salah, María J. Santofimia, Steinunn Gróa Sigurðardóttir, Susanna Spinsante, Lăcrămioara Stoicu-Tivadar, Hilda Tellioğlu, Andrej Žgank, Hazim Kemal EkenelIntelligent 2025 · DOI · OpenAlex

Abstract

Active assisted living (AAL) aims to use innovative technologies to create supportive, inclusive, and empowering applications and environments that enable older, impaired or frail people to live independently and stay active longer in society. This paper provides an introduction to the main concepts of AAL, provides a brief history of the evolution of such technologies, and gives a functional view of how AAL system architecture can be conceptualized. It then provides a taxonomy of AAL technologies and applications, followed by technological underpinnings of audio- and video-based AAL.

Introduction: The Challenge of the Silver Economy

As the global population ages, the "silver economy" faces a critical challenge: how to help older and frail individuals live independently for longer without overwhelming existing healthcare infrastructures. While the field of Active Assisted Living (AAL) offers a solution through Information and Communication Technologies (ICT), the research landscape has historically been fragmented. This paper synthesizes decades of research into a coherent framework, addressing the lack of standardized definitions and taxonomies while highlighting the tension between the need for powerful data—such as video and audio—and the fundamental right to privacy within the home.

The Four Generations of AAL

The evolution of AAL technology is best understood through four distinct generations, marking a transition from reactive, manual devices to proactive, AI-integrated ecosystems.

The four generations of AAL technology, showing the transition toward AI-based functionalities and ethical guidelines.
  1. 1st Generation: Characterized by wearable devices for safety purposes, such as emergency pendants. These systems were reactive, requiring user contribution and response via call centers.
  2. 2nd Generation: Introduced home sensors and textile-heavy devices. These systems utilized automatic data analysis to detect hazards and send alerts, though they often felt intrusive and hindered daily activities.
  3. 3rd Generation: Focused on the seamless integration of home and wearable devices. These systems provided risk detection and smart assistance but often lacked deep consideration of privacy and user acceptance.
  4. 4th Generation: The current frontier, defined by advanced AI and robotics. This generation prioritizes "Ethical Excellence," utilizing co-design approaches and ensuring privacy and legal issues are considered at the fundamental technological level.

Functional Architecture and Taxonomy

To move beyond fragmented solutions, a unified three-layer functional model is proposed for AAL systems:

  • Sensing/Perception Layer: This layer utilizes the Internet of Things (IoT), wearables, and environmental sensors (e.g., thermal, radar, and microphones) to gather raw data.
  • Data Interpretation Layer: Here, Machine Learning (such as SVM and k-NN) and Deep Learning (CNNs) are employed to recognize complex activities of daily living (ADLs) or detect emergency events like falls.
  • Actuator/Interaction Layer: This layer provides the necessary feedback to the user or caregiver via user interfaces, social robots, or smart home appliances.

This architecture is supported by a detailed taxonomy that classifies AAL systems based on their technological underpinnings (how they function) and their assistive scope (who they help and what specific needs they address).

The Privacy-Utility Trade-off

A central theme in modern AAL is the balance between sensor utility and personal privacy. While video and audio sensors provide the most informative data, they are also the most intrusive. The paper champions "privacy-aware" sensing. For example, low-resolution thermal arrays (e.g., 32x24 pixels) or radar can effectively detect falls and room occupancy without revealing the identity or specific details of the individual.

Furthermore, the field is shifting from "handcrafted" features—where engineers manually define the parameters of a fall—to representation learning. Deep Learning models, particularly Convolutional Neural Networks (CNNs), are now capable of autonomously identifying relevant patterns in raw data, significantly improving accuracy in gait analysis and vital sign monitoring.

Key Results and Performance Metrics

Quantitative findings in the literature support the efficacy of these advanced sensing methods:

  • Fall Detection: Radar-based systems have achieved sensitivity and specificity rates exceeding 97% and 90%, respectively, offering a high-performance, camera-free alternative.
  • Thermal Efficacy: Even extremely low-resolution IR arrays (as small as 8x8 pixels) have proven effective for human detection and counting.
  • Deep Learning Dominance: DL methods consistently outperform traditional ML in perception tasks, including the ability to monitor heart rates through color change amplification in video data.

Limitations and Future Directions

Despite these advancements, several hurdles remain. The "Black Box" nature of Deep Learning necessitates the development of Explainable AI (XAI) so that clinicians and users can understand why an alarm was triggered. Additionally, legal frameworks like GDPR are often struggling to keep pace with "lifelogging" technologies.

There is also a critical shortage of high-quality, annotated, real-world datasets, particularly for audio-based AAL in noisy domestic environments. Finally, researchers warn of a growing "access gap," where these life-extending technologies may only be available in high-income countries, potentially exacerbating global health inequalities. The future of AAL depends on bridging these gaps through ethical design and standardized, cross-border regulation.

Summary updated

Abstract

Abstract Active assisted living (AAL) aims to use innovative technologies to create supportive, inclusive, and empowering applications and environments that enable older, impaired or frail people to live independently and stay active longer in society. This chapter provides an introduction to the main concepts of AAL, provides a brief history of the evolution of such technologies, and gives a functional view of how AAL system architecture can be conceptualized. It then provides a taxonomy of AAL technologies and applications, followed by technological underpinnings of audio- and video- based AAL.
BibTeX
@inproceedings{colantonio2025historical,
  author = {Sara Colantonio and Slaviša Aleksić and Jean Calleja‐Agius and Kenneth P. Camilleri and Anto Čartolovni and Pau Climent-Pérez and Stefania Cristina and Vladimir Despotović and Hazım Kemal Ekenel and Mustafa Ekrem Erakın and Francisco Flórez-Revuelta and Danila Germanese and Nicole Grech and Murat Emirzeoğlu and Ivo Iliev and Anna Sigríður Íslind and Mladjan Jovanovic and Martin Kampel and William D. Kearns and Andrzej Klimczuk and Lambros Lambrinos and Jennifer Lumetzberger and Wiktor Mucha and Sophie Noiret and Zada Pajalić and Rodrigo Pérez‐Rodríguez and Galidiya Petrova and Sintija Petrovica-Klavina and Peter Počta and Angelica Poli and Mara Romanovska and Albert Ali Salah and María J. Santofimia and Steinunn Gróa Sigurðardóttir and Susanna Spinsante and Lăcrămioara Stoicu-Tivadar and Hilda Tellioğlu and Andrej Žgank and Hazim Kemal Ekenel},
  title = {{A Historical View of Active Assisted Living}},
  booktitle = {Intelligent},
  year = {2025},
  doi = {10.1007/978-3-031-84158-3\_1},
}
Figure