Ambient Learning Spaces: Systemic Learning in Physical-Digital Interactive Spaces

Algorithmic and Aesthetic Literacy. Emerging Transdisciplinary Explorations for the Digital Age. Opladen ; Berlin ; Toronto : Verlag Barbara Budrich 2021, S. 97-115
Padagogische Teildisziplin: Medienpadagogik;


Introduction
We developed and discuss here a new system concept for digitally-enabled teaching and learning environments that include classical electronic media concepts and enables a pedagogy that is only possible through the ubiq uitous dynamic new digital media.The basic concepts for our Ambient Learning Spaces (ALS) environment have been based on contemporary ped agogical theories, such as those described in Assisted Learning by Arnold (2011) or Expansive Learning by Engeström (2014) taking into account that learning processes are becoming less dependent on teaching content and more on the active process of learning itself.Together with the notion of a ubiquitous and pervasive digital environment, as discussed by Weiser (1991), self-driven and creative learning supported by such a digitally en riched, ubiquitous learning environment has a high potential for future education.Technical, methodological, as well as social competencies need to be considered together as they influence the individual development of learners to a high degree.We have to think about a learning culture that favors forms of self-organized and self-directed learning using generally available ubiquitous digital extensions of ourselves in the sense of McLuhan's Extensions of [Hu] Man (1964).Learning has to be oriented towards a living learning process in which learners discover and construct new and different knowledge structures and content.Especially in highly digital and therefore increasingly virtual environments, joint learning and references to the real world seem to be important for grounding and progress.This implies the need for body-and space-related digital learning.As the human body and human perceptional abilities, as well as the sensomotorical skills are the foundation for aesthetic processes, ALS is able to naturally serve as a platform to foster aesthetic and creative learning.

Methodology
The concept and implementation of Ambient Learning Spaces or short ALS is based on several methodological foundations.In this section, we will refer to some of the more important pedagogical theories as well as the re sulting systemic model of ALS.

Pedagogical Foundations of Ambient Learning Spaces
At the end of the 20th century and now in the 21st, pedagogy and didactics mostly follow interactionist-constructivist (e.g., Reich 2007) and systemicconstructivist approaches (e.g., Arnold 2011).Contemporary didactic ap proaches assume that learning is an active construction process, where a learner creates an individual aesthetic and algorithmic (functional) repre sentation of the world.However, this process is not purely subjective, as early constructivists like Maturana and Varela (1992) hypothesized since each subject is in relationships with other subjects within different commu nication communities.Via communication systems, any statement about reality is subject to viability, because of changes in interests, power possibil ities, as well as social, economic, cultural, symbolic capital formations in the sense of Bourdieu (1991).Learning thus depends strongly on individual prior knowledge and the social, natural and technical environment, in which learning takes place.Individual aesthetic expression and communi cation will drive creative co-construction processes.
Even when we have to question Piaget's rigid step-by-step development model (1977) because of the social, societal and cultural influences of learn ing, his fundamental approach to learning theory remains groundbreaking: It is not possible to transfer knowledge from one person to another person; instead, each person must construct it by himor herself depending on his or her previous knowledge and skill level, attitudes contextualized by chal lenges and learning contexts.As a result, learning is not passive storage, but active and creative construction of knowledge, which should be supported by learning environments.In contemporary constructivist models, the role of a teacher is not to impart knowledge, but to support learners in their individual learning processes through a balanced measure of instruction.The learners shall independently deal with learning content, for example by content selection, discovery of relations, and algorithmic combination of chunks of already available knowledge.
Comparable to the interactionist-and systemic-constructivist ap proaches, but not as prominent, is the theory of Expansive Learning in the sense of Engestrom.His pedagogical approaches follow the so-called cul tural-historical theory of activity, founded in the 1920s by researchers such as Vygotskij (2012) and Leont'ev (1978) and further differentiated in Criti cal Psychology for Self-Determined Learning by Holzkamp, as discussed by Engestrom (2014).According to Critical Psychology, learning in general means the appropriation of an object meaning by a learning subject and not the achievement of a normative educational ideal.In addition to concrete things, this also includes abstract and symbolic references.Thus, Expansive Learning addresses individual or collective learning processes with the goal of extending action possibilities, competencies, and self-determination within processes of co-construction.
This means for the design of digital systems supporting learning by fol lowing contemporary pedagogical-didactic models as briefly presented above, digital learning systems and environments shall be enabling and sup porting the individual construction of sustainable knowledge, i.e. knowl edge that finds relevance of use in the reality of daily life.In this case, these systems support individual and also cooperative learning by establishing a strong relationship to the physical world connected to the human body as well as the social reality of the learners.In a recent development of inter active multimedia systems, we see a revolution in the design of user inter faces like in peripheral media, tangible media, mobile media or wearables (Fig. 1) embedding the human learner not only mentally, but even bodily.Learning software, which uses such capabilities of new interactive inter faces, has high potential to support creative learning processes and collab oration as expressions of aesthetic co-construction, and, above all, a rele vant relationship to life physically and mentally.

Requirements for a Systemic Media Platform for Contemporary Pedagogy
To support constructivist pedagogical concepts, including the world of cur rent digital media, we need a systemic technological concept for a digital media platform that can be connected to real world contexts during teach ing and learning.The platform has to serve as the technological substrate of digital aesthetics as well as algorithmic teaching and learning inside and outside the classroom.Media created or provided need to be accessible by a variety of inter active frontend applications by the learners for stationary and mobile usage and have to be re-usable and re-combinable in different teaching and learn ing contexts.Access to the media is possible through user interfaces that permit different interaction forms, like image and video viewers, interactive form filling, algorithmic image and video graphs, as well as augmented and virtual realities or even wearables.
A technological multimedia platform for teaching and learning needs to support a wide variety of media types like text, sound, image, video and 3D objects.Media shall be constructible as a structured collection of basic me dia for different input and output devices.The platform needs to be able to dynamically generate device dependent variations of source media like im ages, videos or 3D objects with a specific format or resolution as needed.For example, mobile smartphones will be provided with lower resolution images than large public screens.The conversion of media shall be per formed in automatic processes.The platform shall provide flexible per sonal, group and public ownerships for media objects or collections of them.
The media created need to be enriched by informal annotations refer ring to the application contexts.Additionally, they shall have relations to higher level concepts like taxonomies referencing to the global DBpedia or self-constructed taxonomies of knowledge.These formal relationships to gether with the informal tags will transform the media into semantic media related to real world and real usage.Tags usually stem from ad hoc refer ences during social use.Certain tagging shall be done semi-automatically from the context of use itself, like the project ongoing or geographic and chronological references throughout the process of use.
The media need to be hosted and organized in a cloud-based distributed media database to be accessible anytime and anyplace for contextualized learning and teaching.

The ALS Architecture and Applications
In over 10 years of research we developed a system with capabilities de scribed in the previous section.The system has been called Ambient Learn ing Spaces (ALS) (Winkler et al. 2011, Herczeg et al. 2019).The pedagogical requirements described in the previous section have been translated into a technical ALS system architecture (Herczeg et al. 2020a, b).

The ALS Learning and Teaching Applications
The ALS environment provides an increasing number of different applica tion modules (Fig. 2) used by learners and teachers to initiate and organize learning processes.These hardware and software modules can be used to construct and enrich specific learning contexts.All of these frontend mod ules use the same backend and middleware system called The Network Environment for Multimedia Objects (NEMO) (Feldner et al. 2009, Lob et al. 2010) with common backend features and a shared storage layer for all mul timedia objects generated and used during the learning processes.In the following subsections, some of the most important and most advanced modules will be outlined and shown.New modules can be added dynami cally to the flexible and open ALS platform.As a central integrative element, built as a stationary frontend for class rooms or school foyers, we implemented a system called the Interactive Wall (IW).The IW is a social entry and access point for teachers and learners and supports cooperative learning (Winkler et al. 2017).It consists of an arbitrary number of large interactive displays that can be installed in build ings and rooms (Fig. 3).The displays consist of multiple affordable TV screens which are equipped with touch-sensitive frames.The large displays are connected to a distributed array of client computers, which themselves are connected to NEMO.Most of the ALS learning applications can be ac cessed or controlled through an IW.

MediaGallery
The MediaGallery is an application running on the IW.As the IW is provid ing public interactive displays, the MediaGallery enables learners to explore personal, group-based or public media collections about their learning topics (Fig. 4).Each collection consists of a set of images, podcasts, videos, docu ments or 3D objects that are either manually combined or selected by semantical correlations and search.When multiple media galleries are available, the IW has an area to place randomly selected dynamic previews of public media collections on the IW.Media collections display, connect and separate content.They represent ownership and topical common ground.Students explore information elements along with semantic relationships provided by the global DBpedia database.Ontologies are leading to knowl edge chunks related to each other in a semantic mesh (Fig. 6).While being used, the system is constantly searching the World Wide Web for new enti ties related to the already displayed and selected ones.Different algorithms to calculate similarities and differences of such entities help to choose from a large number of candidates found to be displayed.SemCor gives insights to very large information repositories and supports associative learning based on serendipity as well as student' s own curiosity (Herczeg et al. 2020b).
Figure 6.SemCor interacting with knowledge entities and relations (Screenshot: own work)

MoLES
MoLES is a web application for smartphones and tablets to support taskoriented teaching mainly outside school for example in urban space, natural habitats or museums (Fig. 7).Teachers or more advanced students can de fine structured tasks to be solved within project-oriented teaching (Winkler et al. 2009).The students will work on tasks like challenges, which they have to solve mainly outside school.While completing the tasks they create and collect media entities captured with their mobiles.These media will auto matically be uploaded into the NEMO system and become available for fur ther use as a documentation of results in a project-centered media gallery (Herczeg et al. 2020a).

InfoGrid
InfoGrid is a mobile learning app that uses Augmented Reality (AR) tech nology to display personalized and contextualized information such as im ages, videos, interactive 3D objects as virtual overlays for physical space (Ohlei et al. 2018a, b, Herczeg et al. 2020a).InfoGrid can be used in places where the image targets that trigger the augmentations are available.When the app is started, it shows the physical environment on the mobile screen along with overlaid virtual information (Fig. 8).Students can, for instance, use InfoGrid to create augmented posters for their projects, set up exhibi tions of physical artifacts or create AR games inside or outside school.

The ALS Portal and the ALS Authoring Systems
For a learner-centered self-driven learning process it is a basic requirement that students and teachers themselves are capable of creating and annotat ing media and their aesthetic expression as well as managing the platform by defining users or by activating or deactivating certain applications.Therefore, we developed a web-based management module called ALS Por tal, which provides the above mentioned functionalities.To enable teachers and students to manage their created media we also developed several authoring tools to work with images, videos, and 3D objects.Furthermore, we created tools to support the placement of virtual objects in physical set tings for AR applications.

ALS Portal
The ALS Portal can be used to create and edit information for all ALS learn ing applications and allows activating or deactivating these applications.Users can access the ALS Portal through a regular web browser or the IW.
After logging into the ALS Portal, a list of ALS applications will be pre sented, which are available for the current user (Fig. 9).The user can then create or edit information.All data entered through the ALS Portal will be stored inside the central NEMO backend.

VideoEdit
VideoEdit is an easy to use web-based tool that can be used to create and edit video footage.Users can upload their media files such as images and videos either through their mobile phones or through a local PC.These media files can then be merged into a new video file.VideoEdit supports adding a separate audio track as well as a text overlay track and has addi tional functions such as increasing and decreasing the volume of the result ing video.After preparing a video, NEMO renders the video using the FFMPEG framework.The resulting MP4-file is automatically made avail able for ALS applications.

3D Converter and Editor
We implemented a special functionality called NEMO Converter 3D (NOC3D) for the automatic generation of 3D objects from 2D video and photographic footage (Bouck-Standen et al. 2017, 2018a).The user can up load videos or photos into NOC3D, which automatically processes these files using photometric methods.After completing the process, the object converter delivers a 3D model stored in NEMO.The resulting 3D models usually contain unwanted artifacts and are randomly placed in space.3DEdit can then be used within the web-based ALS Portal to remove unwanted artifacts from the 3D-Object and to change position, scale and orientation of the 3D object in space.After finishing this process, these ob jects can be used in other ALS applications such as the MediaGallery or InfoGrid.

Narrator
With the Narrator Module of NEMO it is possible to create storylines for ALS applications.The user can create and annotate story elements with semantic information in a way that the NEMO framework can create rela tions between the elements (Bouck-Standen et al. 2018b).To experience the storylines created, for example the InfoGrid AR application can be used.The Narrator will guide the user depending on their topics of interest or earlier visits along a path of activities.The dramaturgical algorithmic struc ture can be created based on some classical models of drama like such from Aristotle or more contemporary ones.

Profiling and Personalization
All media stored in NEMO using the ALS Portal can be semantically anno tated during the creation and uploading process.Besides topical areas, the annotation can also include information about the language used within the media or the recommended age of the target users.Using the profiling and personalization service all ALS applications can select the appropriate data for the user (Bouck-Standen et al. 2018c).ALS has been used in different ways for language teaching and learning.The NEMO framework also in cludes a converter service so that all media elements are converted into the correct format depending on the currently used device that requests the in formation.

The NEMO Services
The ALS architecture contains a Service Layer to support application mod ules with several services.

User Authentication
The Administrator of the ALS Portal can setup institutions (e.g.schools) and assign teacher accounts to a specific institution.With their accounts the teachers themselves can then create and edit accounts for the students (Bouck-Standen et al. 2018c).The teachers can also define student work (project) groups.Depending on the ALS application used, the teacher has the role of a moderator with rights to publish and unpublish information for public display or group use.

Tracking
The NEMO framework includes a Tracking Module to anonymously log and track user requests from all ALS applications.To inspect the tracking data the ALS Portal contains a map-based tracking visualization module.This supports insights on preferences and possible obstructions.It shows statis tics of applications used or media queries indicating how often media files are requested.The tracking function will help to optimize didactic con cepts.

Cognitive Services
When the ALS Portal is used to upload media files into NEMO for any ALS application, it is necessary to provide semantic tags.To simplify and speed up the process of tagging, Cognitive Services analyze the selected media and provide suggestions for the user.The user can accept or reject the sug gestions for tagging.The Cognitive Services layer has used artificial-intelli gence-based methods for classification.

The ALS Logic and Database
The backend of the ALS architecture is represented by a Logic and Database Layer to provide persistent semantic storage of media.

Central Logic
The ALS Central Logic of NEMO contains mechanisms to store and re trieve media content depending on applications and frontend devices as well as ownerships of media.It connects the NEMO Services with the data base.
Semantic Database NEMO makes use of a semantic database to store information in an RDF information model.The open source BrightstarDB has been used for the implementation.The SPARQL query language can be applied by the Cen tral Logic to access content in the database.Information will be valorized into knowledge-based structures referring to meaning in the real world.
Cloud-based Networking NEMO can be hosted on any physical or virtual machine in a network.Sev eral NEMO instances may be connected and layered to provide content independent of its current location.NEMO can be seen as a distributed cloud-based storage system for semantic teaching and learning content to serve any frontend environment and define ownership, digital rights and security needs.It can be hosted on a school server to keep all data within the institution.

Evaluation Studies and Results
During the last years the ALS Framework and its applications have been evaluated in many different dimensions and criteria in the lab, inside and outside school as well as in museums.

Pedagogical Models and Studies
The usage of the ALS System in school teaching has always been accompa nied by studies.Apart from different qualitative surveys, student's activities on the IW are tracked anonymously via NEMO providing detailed statis tical data.Teachers and other authorized persons can check how often the applications are used inside the ALS Portal.Furthermore, the access of the media themselves and the media volumes created over time can be ana lyzed.This allows discussing the practical values of the applications with respect to their pedagogical relevance and effort by the teachers themselves.
ALS applications have been used so far by several hundred teachers in more than 20 different schools on all class levels to construct their own understanding of the digital transformation in school.Every teacher has personal didactic methods and pedagogical values.It would be of no use to evaluate the ALS environment pedagogically as a fixed solution indepen dent of the special needs, constructs and contexts of usage.However, the modules have proven to be helpful in a variety of teaching contexts over school terms.In many cases these contexts have been characterized by aes thetic-creative processes based on information modeling with aesthetic digital media and their appreciation.
We have been providing seminars for teachers to train the use of differ ent applications technically as well as didactically through the German Institute for Quality Development at Schools Schleswig-Holstein (IQSH).

Usability and User Experience
The frontend applications have been developed in user-centered develop ment processes under participation of many teachers and students.Most of them have been evaluated for their usability with SUS or ISONORM ques tionnaires.The formative evaluation results have been used to improve the aesthetic design quality besides the usability and user experience of the functionality through many development cycles.Most of the applications show a high level of usability.However, being a living platform there are always new applications, which are still in an early stage of development and evaluation.

Technical Validation and Performance
The ALS environment has been implemented as a layered .NET-based ar chitecture, which supports agile development.Most applications have been programmed in C#.Besides SQL modules, the database layer has been built with the BrightstarDB framework.Frontends have been implemented mostly for web browsers in HTML5 and CSS3 or as native apps for iOS and Android for mobiles.
The ALS system is a large hardware and software system that has been tested in many ways to deliver the performance that is needed inside and outside school.The current system is able to host teaching projects over years in schools and museums with thousands of media elements stored and managed.As a result of the database concept with a separation of struc tural information and large media objects, the system is highly scalable in volume.
ALS is a running environment that is currently used on a daily basis by several schools and museums.However, being still a prototypical research environment with a limited operational dimension, ALS can at least serve as a blueprint for future digital platforms for educational use in general.

Conclusions
Ambient Learning Spaces (ALS) is a large prototypical teaching and learn ing environment for a wide variety of learning contexts with aesthetic and creative learning processes at school.ALS has been based on contemporary learning theories like Assisted or Expansive Learning bringing the learners into an active constructive role.
A semantic media repository allows the use and reuse of media in dif ferent contexts and for different interaction devices.A suite of modular learning applications for stationary and mobile learning has been built, applied and evaluated in real teaching contexts.The usability tests and ped agogical case studies show that ALS applications can be used effectively and efficiently by students and teachers.
ALS support body-and space-related learning by providing a large vari ety of frontend systems from wearables through mobiles to room-based installations.It addresses directly aesthetic processes in perception and action.The applications are modular and the media reusable to enable the teachers to adapt and reuse the ALS system and their media content accor ding to teaching needs and allow creative iterations within and across groups of learners.ALS are ubiquitous and pervasive and do not imply or force didactical methods.It has been made available to a large number of schools and museums over several years.
ALS show the transition from computers used as digital media to a digi tal knowledge platform for constructive teaching and self-directed aesthetic and algorithmic learning.

Figure 1 .
Figure 1.Shell model of ubiquitous media used in ALS.(Own work)

Figure 3 .
Figure 3.The Interactive Wall (IW) as a social access point to ALS. (Photo: own work)

Figure 4 .
Figure 4.A sample media collection in the MediaGallery.(Screenshot: own work)

Figure 5 .
Figure 5. TimeLine to interact with chronological knowledge.(Screenshot: own work)

Figure 9 .
Figure 9. ALS Portal for management of ALS applications and content.(Screenshot: own work)