A public selection procedure is called, based on qualifications and interview, for the assignment of n. 1 research fellowship grant for collaboration in research activities.
Title of the research: Design and implementation of techniques for automatic 360-degree video analysis and processing
Description: The research activities will aim to use/devise innovative techniques of image processing and computer vision for the design, implementation and validation of 360-degree video analysis tools. Of particular interest will be the techniques concerning the re-identification and tracking of people to model their interests, behaviors and reactions to the suggestions provided by a recommendation system.
Supervisors: Prof. Liliana Lo Presti and Prof. Marco La Cascia
Duration: 12 months
The total gross amount for 12 months is equal to € 26.525,00 including the charges borne by the Administration and will be paid to the beneficiary in deferred monthly instalments.
General admission requirements: - Italian Master‘s Degree, of a single-cycle Degree or equivalent qualifications issued by foreign countries
- a professional scientific curriculum suitable for carrying out research activities in computer vision
Citizens of EU Member States and non-EU countries must also possess the following requisites:
- enjoy full civil and political rights in the states of belonging or provenance;
- be in possession, with the exception of the ownership of Italian citizenship, of all other requisites envisaged for Italian citizens;
- have a working knowledge of the English language
Application and deadline: - The application for admission to the selection procedure, drawn up in free paper according to the form attached, must be sent electronically (please, see attached files) with the candidate's CV and publications (scientific articles).
- For the participation in the competition candidates are required, under penalty of exclusion, to pay within the deadline for submission of the application, a contribution for organizational costs in the amount of € 50.00.
- Deadline: 27 February 2022
Link: https://titulus-unipa.cineca. it/albo/viewer?view=files%2F004435328-UNPACLE-55466aa9-badc-4d1d-8803-69bd62c6d0b7-001.pdf
Link Form: https://www.unipa.it/amministrazione/direzionegenerale/serviziospecialericercadiateneo/u.o.assegnidiricerca/. content/documenti/2021/application-form-and-annex_--nuovo.dochttps://www.unipa.it/amministrazione/direzionegenerale/serviziospecialericercadiateneo/u.o.assegnidiricerca/. content/documenti/2021/application-form-and-annex_--nuovo.doc
Dear all, We have a postdoc opening on developing Arabic handwritten text recognition systems at the NC State University, North Carolina, USA.
No Arabic knowledge is needed but please make sure that you are good at deep learning and you are in one of the following fields: document analysis, computer vision, and image processing. You will be jointly supervised by a humanity faculty member and an ECE faculty member (me).
More details about the opening including the application link can be found here: https://jobs.ncsu.edu/postings/152789
If you have any questions, please don't hesitate to let me know. I also highly appreciate it if you could forward this opening to relevant parties.
Sincerely, Chau-Wai Wong Assistant Professor
Electrical and Computer Engineering NC State University, Raleigh, North Carolina, USA https://people.engr.ncsu.edu/cwong9/
The LITIS laboratory at the University of Rouen is looking for a candidate for a 3-year PhD, funded by the ANR (National Agency for Research), around the design and prototyping of a bio-inspired vision system for assisting indoor and outdoor mobility of visually impaired people (VIP).
Contact : [email protected], [email protected]
Start : as soon as possible (April 2022 latest)
Keywords: ========= navigation (indoors and outdoors) and its bio-inspired foundations, computer vision, localization and tracking, multimodal data and their fusion, mobility assistance of visually impaired people (VIP).
Scientific program: The national project “SAM-Guide: Spatial Awareness from Multimodal guidance” (2021-2025) brings together computer science research teams (LITIS/URN, CerreV (Centre de Recherche Risques & Vulnerabilities /UCN, GIPSA-Lab/UGA , LPNC/UGA, and CMAP/Ecole Polytechnique) around the navigation indoor and outdoor of VIP. The project aims to rethink travel assistance in an " inclusive" approach, i.e. suitable for sighted people and VIP.
The targeted approach could be based on (1) the audio-tactile maps displayed on an original haptic device under development at Rouen University, and (2) on the navigation of mammals (specific cells of the hippocampus).
The PhD candidate will be responsible for proposing the model of such mobility, defining multimodal data acquisition and fusion protocols (vision, IMU, Galileo, etc.), and designing, prototyping and validating (in particular with end users) different vision and data fusion algorithms constituting the technological base of the project.
Skills required: ============ The candidate must have a Master level degree with skills in vision (for robotics), robotics (e.g. navigation, design and prototyping of mechatronic systems such as force feedback systems), image processing and computer engineering (programming, algorithms).
Apply. ========= The application must be sent by email, including a detailed CV, a cover letter, transcript of records (masters 1 & 2) and the names of two people to contact (from the master's team). An (e-)interview will be offered if the candidate's profile appears to meet expectations.
Please include in your application within one pdf-file sent by e-mail to [email protected], [email protected]
Work environment: ============== The PhD, funded by the ANR SAM-Guide project, will take place in the LITIS laboratory, located on the Madrillet campus of the University of Rouen Normandy, France. The doctoral student will integrate the STI (Intelligent Transport System) team, which constitutes an international multidisciplinary environment allowing interaction and collaboration with recognized experts in the fields underlying the ANR project.
For more information do not hesitate to contact us.
Internship position: Human-in-the-loop Multi-Agent Reinforcement
Learning for Self-Adaptive Energy Systems
Context & objectives Many problems in energy management cannot be solved by individual components or systems working in isolation because they require different expertise, resources, and information. Energy systems are, therefore, commonly composed of interacting components and subsystems, owned by different actors and stakeholders with their individual objectives, constraints, and policies.
Due to historical, political, physical, or social constraints, it may be impractical or undesirable to permanently synthesise the necessary components into a single entity, thus cooperation among the independent components is required. This implies decentralised decision making and control.
Multi-agent systems (MAS) provide an efficient approach to model and design such systems, enabling the loosely coupled components represented by agents to interact to achieve their individual goals as well as system’s objectives [1-4]. Dynamics and uncertainty present an integral challenge. The variability of the systems and the effects of the decisions made by different agents make it impractical, if not impossible, to determine complete strategies at design time. Often, it is necessary to adopt some sort of learning capability to improve the strategy at run time.
Systems need to adapt themselves (e.g., their decision making) as they observe and learn about the environment. Reinforcement learning allows agents to explore the unknown environment and to improve their behaviour through trial-and-error interaction with their dynamic environment. Reinforcement learning focuses on single-agent learning. Its extension in a multi-agent context considering multiple learning agents is known as multi-agent reinforcement learning (MARL) [5-8]. In principle, learning agents are designed to learn autonomously based on their reward function and learning algorithm.
However, fully autonomous learning is often infeasible in the presence of highly complex problems or potentially dangerous outcome, in which case interactive guidance from humans is necessary. Human in the loop learning (HILL) addresses the challenge by involving humans in the learning process of the agents where humans can intervene and give advice [9-12].
The mission of this internship is to investigate the potentials of employing MARL and HILL in MAS to design self-adaptive energy systems that can learn to adapt themselves in response to dynamics. The proposed solutions will be applied in the management of local energy communitiesthat are equipped with renewable energy sources(RES).
More precisely, we aim at increasing RES penetration and electricity cost saving by optimising energy consumption to improve self-consumption in the communities and to support grid constraints (e.g., through demand response schemes), while ensuring user comfort. MARL and HILL will be employed to learn user preferences from data or observations and to consider dynamic changes in the system, which aids agents in their decision making. Various on-going projects in the lab on energy communities could provide further specifics on the use-cases and data for this internship.
Work environment ENGIE Lab CRIGEN is the corporate research centre of the ENGIE Group dedicated to new gases, new uses of energy, digital solutions, and emerging technologies. Located in the Paris region at Stains (93), it has 200 employees. It provides tested, proven, and marketable industrial applications, as well as new offers based on the development and pooling of innovative ideas, scientific knowledge, and technical expertise.
Its ability to innovate is a key advantage for the ENGIE Group. A part of CRIGEN, the CSAI Lab (Computer Science & Artificial Intelligence) carries out work in the fields of Computer Science and Artificial Intelligence (Machine Learning), Interoperability for Digital Ecosystem, and Distributed Autonomous System. The intern will be integrated into CSAI and work closely with scientists in machine learning and MAS.
Tasks and responsibilities • Prepare a state of the art on MARL, HILL, and their applicationsin energy management
• Propose relevant strategies to integrate MARL and HILL into MAS to improve energy management (e.g., learn user preferences and learn to improve decision making)
• Design and implement a proof of concept to validate the proposed solutions
• Present and report findings orally and in writing
• Contribute to scientific papers
Education: Bac+5 (Master 2 or final year of engineering school) in Computer Science or related fields
Requirements • Knowledge in machine learning and deep learning; experiences in reinforcement learning or HILL is a plus;
• Knowledge in MAS such as agent models and architectures, coordination, and agent- oriented programming
• Experiences in performing innovative research in applied machine learning and AI
• Fluency in English (spoken and written)
• Pro-active, independent, and comfortable working in teams
• Knowledge in optimisation and energy domain is appreciated.
Technical skills
• Advanced skills in Java and Python
• Software development (analysis, design, implementation, test) and Git
• Knowledge and skills in machine learning frameworks (scikit-learn, TensorFlow, or PyTorch)
• Experiences with agent platform (e.g., JADE, JaCaMo, ...) is a plus. Work conditions
• Location: ENGIE Lab CRIGEN, 4 Rue Joséphine Baker, 93240 Stains (Paris region, RER D, station Pierrefitte - Stains)
• Duration: up to 6 months • Start date: as soon as possible
Application: Interested applicants may send their CV, cover letter, and transcripts of the last
2 years to the following people:
Sarra BEN ABBES: [email protected]
Rim HANTACH: [email protected]
Oudom KEM: [email protected]
Hello Would it be possible to post this advertisement for a position as a teacher-researcher at the School of Air and Space on your mailing list and in the advertisements
The Air and Space School is recruiting a teacher-researcher specializing in cybersecurity of cyber-physical systems through secondment or on a fixed-term contract. Coming from the computer science or industrial computer communities, the person recruited will provide teaching to air force and space officers in initial or continuing training in the various bachelor's or engineering courses and specialized masters carried by the school.
https://www.ecole-air-espace. en/wp-content/uploads/2022/02/Teacher-Researcher-in-Cyber-Security-of-Cyber-Physical-Systems.pdf
thanks in advance
Prof. Eric BLANCO School of Air and Space Director General of Education and Research
French Air Force Academy Vice Rector for education and research
tel : +33 413938373
TRUNK: 8647018373
mobile +33 652553886
Body: Research Engineer
Quota: 100%
Desired start date: 05/01/2022
Position open to: CDD contract workers ❑ CDI contract workers ❑ Permanent staff (civil servants)
Duration of the contract (if open to contract workers): 12 months
Assignment of the post Geographical location SMAC team Computer Science Research Institute of Toulouse
University of Toulouse III – Paul Sabatier 118 route de Narbonne 31062 Toulouse Cedex 09
Context and work environment Heiress of the former University of Toulouse founded in 1229, Université Toulouse III - Paul Sabatier (http://www.univ-tlse3.fr) is a higher education and research establishment founded in 1969 the merger of the Faculties of Medicine, Pharmacy and Science.
It ranks today among the first French universities by its scientific influence, the diversity of its laboratories and the training it offers in science, health, sport, technology and engineering.
With a budget of 400 M€, it is strong of more than 4300 staff including 2500 teaching staff and/or research, has 69 research structures (including 41 joint research units), hosts over of 35,000 students and has 7 components (Faculty of Science and Engineering, two Faculties of Medicine, Faculty of Odontology, Faculty of Pharmaceutical Sciences, Faculty Sciences of Sport and Human Movement and two IUTs and the Midi-Pyrénées Observatory).
She is established in 4 departments, 8 cities and is spread over 11 sites for an area of 264 hectares.
Job Information For more information on the position, do not hesitate to contact Mrs. Marie-Pierre Gleizes: Marie- [email protected])
Salary: civil service salary grid (2038€ gross per month for a beginner IR) Sending applications
CV and motivation letter to be sent exclusively to the following address:
Development of a customizable configuration approach for the digital work environment
Job: Post doctorate
Salary: 2600-2700 € monthly gross income
Starting Date: April-May 2022
Duration: 12 months
Location: CNAM, Paris (75) and IRSN, Fontenay aux Roses (92)
Context: Conservatoire National des Arts et Métiers (CNAM, a public high education and research institution) and Institut de Radioprotection et de Sûreté Nucléaire (IRSN, i.e. the public expert in research and expertise relating to radiological risks, the protection of man and environment) have joined forces in a joint «Cartographie Immersive Etendue/ Extended Immersive Mapping (CIE)» project to innovate the digital environment for researchers. In this context, we are recruiting a post doctorate whose mission is to propose a design and a concrete implementation (Proof Of Concept) of the CIE project, the whole constituting the first part of this project.
Description : IRSN operates experimental facilities, scientific and technical equipment, metrology resources and specialized mobile intervention resources. These resources are grouped together by «Plateformes Scientifiques et Techniques / Scientific and Technical Platforms (PST)». The objective of the joint «Cartographie Immersive Etendue/ Extended Immersive Mapping (CIE)» project is to set up a digital environment for the optimization and promotion of the scientific and technical means of PSTs for the benefit of laboratory teams and other potential users. The CIE must make it possible to provide customizable access to PSTs and associated services (3D visits, planning, economic monitoring, community networks, etc.) according to several parameters such as the user's profile (internal, external, their profession, its hierarchical position, etc.) and its information / data needs (type, level of detail, mode of restitution).
To achieve these objectives, a situational engineering approach will be used with an identification of reusable components defined in relation to user needs and the context of their use. Each user will have a digital space specific to their needs built on an evolving library of components.
The design, implementation and evolution of the CIE must follow the principles of modularity, reuse and adaptation to the context, as well as to the intentions and needs of the users. In order to provide a suitable and evolving workspace for each user, the UX (User eXperience) [1] [2] and EX (Employee eXperience) [3] [4] approaches applied in the digital environment of IRSN were favored and should make it possible to build and then enrich the mapping of the scientific resources of the PSTs as well as all the associated services.
IRSN's digital environment currently includes: Digital workplace (https://www.jalios.com/): this environment provides the IRSN intranet (including management of internal communities of practice);
Microsoft tools (Sharepoint online, Power BI, Teams, Office…);
3D virtual tour (https://vrtice.com/): this solution developed by a startup facilitates access to content in a 3D environment; SPOT uses this tool for the promotion of facilities but also for training and management;
CRAFT web database (https://www.bdi.fr/fr/plateforme-craft/): the pilot produced by IRSN currently makes it possible to gather all the data from these laboratories and their facilities, then to share them internally, with its partners but also with the general public;
Internal database search engine (https://www.sinequa.com/): this engine is used to address most IRSN databases (scientific, administrative, technical, etc.);
External database search engine AMI (https://www.chapsvision.fr/data/);
Other tools (LEGISWAY for contract management, SAP ERP, etc.).
Post-doctoral research project:
Prepare a state of the art of existing work in the field of UX and EX systems applicable to the context of the CIE,
Consolidate the previously conducted analysis of user needs in order to identify the list of functional and technical requirements,
Propose a taxonomy of description of PST installations and equipment at different levels by highlighting their possible uses,
Formalize the notion of the UX / EX component and identify a set of components for a given perimeter,
Propose an approach for configuring the digital work environment according to user needs and profiles,
Participate in the definition of the target architecture of the digital environment, considering the information system in place and Enterprise Architecture rules, Implementation of the proposed approach for a given scope (production of a POC).
An apprentice will assist the post doctorate in the digital web development of CIE components (creation of application pilots, interfaces in the digital environment of IRSN).
Application : Please send your application to [email protected]. The application must include:
Current resume;
Cover letter;
Thesis report;
Publications related to the candidate's research work;
Possibly one or more letters of recommendation.
Hello We offer a thesis offer within the Models and Knowledge team at the L3i laboratory in La Rochelle.
This thesis aims to analyze complex and heterogeneous data for the adaptation of robotic scenarios with the consideration of all dimensions of interaction.
The thesis will start in October 2022 for a period of three years.
You can find all the information of this offer on the laboratory website.
https://l3i.univ-larochelle.fr/spip.php?action=acceder_document&arg=1677&cle=165473a0a3d6bc24df3c4d7aabfcd6ac658b8a5f&file=pdf/sujet_bertet_mondou_cle81be17.pdf
Deadline : 25 mars
Damien MondouResearch teacherLaboratory L3i | IT
Department+33 5 46 45 82 62 / (Int) 70 96 [email protected]
Laboratory Computer Science, Image and Interaction (L3i) La Rochelle UniversityAvenue Michel Crépeau17042
La Rochelle cedex 1 l3i.univ-larochelle.fr
From CEA we are hiring!
Title: Postdoc/CDD 2 years contract in Applied Machine Learning
Keywords: Data Stream, Machine Learning, Continuous Learning.
Description: The selected candidate will be part of “data streams” team of SID laboratory (CEA). S/he will work within «Confiance.ai» project, a big French collective of unprecedented scope to design and industrialize systems based on trusted artificial intelligence, in which CEA participates. Developed in SID premises, STREAMER is a cutting-edge data stream processing (Complex Event Processing) framework devoted to analyzing sequential data for electrical or industrial systems. In the scope of the project «Confiance.ai», the applicant will support the team in:
1- Studying trustworthy AI challenges that apply to real data stream processing. Such challenges may refer (but are not limited) to:
1.1 Continuous learning models applied to data streams in images.
1.2 Computation of confidence scores on the performance of the similarity measure for a model deployed in production.
1.3 Continuous adaptation of models to changing environments.
2. Developing algorithmic solutions to the aforementioned challenges.
3. Implementing and integrating such solutions within the STREAMER.
4. Collaborate with the project team and with researchers of the SID laboratory.
5 Participation in dissemination activities (as seminars, workshops, conferences, etc). Challenges described above will be decided on a rolling basis by the “scientific comity” of «Confiance.ai» project based on laboratories proposals. Their duration may be variable and adapted along the project.
Qualifications We look for a candidate with: - Experience in R&D and/or PhD. - Background in machine learning.
- Strong programming skills (Java, Python). - French and English proficiency speaking and writing.
- High motivation, self-learning and autonomy, teamwork skills - Familiar with InfluxDB, Kafka or Redis [8] is a plus.
Where: CEA Saclay (DIGITEO Labs), France
Duration: 24 months.
Salary: From 2000 net euros/month depending on the applicant's experience.
Benefits: Free shuttles CEA from/to the laboratory to many points of Île de France. Partial refund of transport card.
Starting date: From 2022.
Contact Send your CV (better with a motivation letter) to Sandra Garcia Rodriguez ([email protected]) with email subject [job application].
A PhD Research Fellow position in NLP promoting privacy available in the Language Technology Group (LTG) of the Department of Computer Science, University of Oslo (UiO), Norway.
The objective of this thesis will be to conduct research at the intersection of NLP and data privacy, and in particular on the development of "adversaries" to evaluate the robustness of text anonymization methods.
The PhD is fully funded for a period of 3 or 4 years, depending on whether the candidate will participate in teaching activities.
For more information, please see the full announcement here: https://www.jobbnorge.no/en/available-jobs/job/219713/phd-research-fellow-in-privacy-enhancing-nlp
The deadline is February 28, 2021.
Do not hesitate to contact me for any further information.
Hello, IRISA - CNRS (Rennes, France) offers several fixed-term positions in Automatic Language Processing and NLP/multimodality:
- engineer (M/F) NLP: augmented journalism, classification, information extraction, recommendation https://bit.ly/3H974Uj
- post-doc or engineer (M/F) NLP/multimodality: text/image analysis , multimodal dives https://bit.ly/34fNdo7
- post-doc (M/F) NLP: hybridization symbolic learning and deep learning, constrained text
https://bit.ly/32LKkL8
IRISA-CNRS (rennes, France) has several open positions (fixed-term contracts) on NLP and NLP/Computer Vision :
- engineer NLP for journalism, classification, information extraction, recommendation https://bit.ly/3H974Uj
- engineer or postdoc NLP/multimodality : text/image analysis, multimodal embedding https://bit.ly/34fNdo7
- postdoc NLP for fake news detection: symbolic/deep learning hybridization, constrained text generation https://bit.ly/32LKkL8
Contact: [email protected]
Hello, IRISA - CNRS (Rennes, France) offers several fixed-term positions in Automatic Language Processing and NLP/multimodality:
- engineer (M/F) NLP: augmented journalism, classification, information extraction , recommendation https://bit.ly/3H974Uj
- post-doc or engineer (M/F) NLP/multimodality: text/image analysis, multimodal dives https://bit.ly/34fNdo7
- post-doc (M/F) NLP: hybridization symbolic learning and deep learning, constrained generation of text https://bit.ly/32LKkL8
IRISA-CNRS (rennes, France) has several open positions (fixed-term contracts) on NLP and NLP/Computer Vision :
- engineer NLP for journalism, classification, information extraction, recommendation https://bit.ly/3H974Uj
- engineer or postdoc NLP/multimodality : text/image analysis, multimodal embedding https://bit.ly/34fNdo7
- postdoc NLP for fake news detection: symbolic/deep learning hybridization, constrained text generation https://bit.ly/32LKkL8
Contact : [email protected]
Thesis title: Virtual and Augmented Reality Interactions for Knowledge Building in the field of Digital Heritage.
Contacts and thesis supervisors: Ronan Querrec (Lab-STICC – ENIB) [email protected] Florent Laroche (LS2N – ECN) [email protected]
Keywords: Interaction, Virtual Reality, Augmented Reality, Digital Heritage, Animated Conversational Agent, Ontology, Meta-Model.
Background: This thesis topic is part of the ANR LabInVirtuo project (2021-2025). This project is located in the field of knowledge engineering, Virtual Reality (VR) and Digital Humanities (HN) on the theme "Human activities versus history and heritage of industrial cultural landscapes". The objectives of the project are to develop and validate: a) virtual laboratories and interdisciplinary research methods SHS-STIC (IA-RV) from demonstrators of strong historical and heritage interest; b) collaborative methods involving both institutional actors (museums, etc.) and non-institutional actors (workers, etc.) on the cultural history and digital preservation of industrial professions; c) scenarios located in the Virtual Environment (EV) dedicated to mediation.
Classically, the acquisition of knowledge relating to a 3D object is carried out through the collection of primary sources (archives, archaeological traces, etc.) and secondary sources (academic reference works), supplemented if necessary by oral surveys of the actors. Our proposal focuses on the acquisition and restitution of knowledge (declarative, procedural and embodied) in an EV constituted from 3D modeling of a material and multisensory historical and patrimonial nature (by multisensory, we mean an environment where several senses are solicited either in terms of bodily context or participating in industrial activity by the information they give) with several user games. A user game is a series of interactions involving two (or more) actors in the "Intelligent Realistic Sensory Virtual Environment" (EVIRS) in a collaborative science approach: i) researchers (historian, archaeologist, anthropologist, ...); (ii) heritage experts/researchers; iii) Non-academic actors/researchers for the acquisition of information on the material and sensory environment, acquisition/capture of know-how; (iv) mediation experts/researchers/local actors. These collaborative sets of actors will make it possible to acquire, share and restore more detailed and relevant knowledge through the scenario in an EVIRS. In other words, it is about improving the elicitation of knowledge.
The hypothesis we defend is twofold: 1) the immersion of users in an EVIRS where he is engaged from a kinematic, gestural and sensory point of view results in: a) an increased recovery of the various procedural, episodic and autobiographical memories that are usually questioned or worked on in the classic context of oral archive productions. ; b) beyond declarative knowledge, the possibility of capturing gestures and embodied knowledge; (c) better memorization of the proposed knowledge; 2) Collaborative VR allows, beyond the classic approaches in SHS, to develop multidisciplinary research methods and innovative mediations with a qualitative and quantitative leap concerning the elicitation and restitution of knowledge and know-how.
Positioning: Today, the use of digital technology to capture our heritage is acquired by all professionals in the field. Projects carried out within institutions make it possible to virtually manipulate heritage objects [Dima et al., 2014] and present them in their historical context [Rae and Edwards, 2016]. But, to our knowledge, there is no system allowing their operation to maintain their use. Yet VR seems to have a significant effect on the acquisition of knowledge about these historical objects [Li and Chang, 2017] and according to [Barreau, 2017] "in the more distant future, the stakes will be at the level of simulation and interaction with virtual humans of the past".
In addition, the ANR ReSeed project (led by the LS2N – www.reseed.fr) demonstrated the interest and the need "to set up a new methodology, a tool and an interoperable format to allow the alliance of the semantic and physical digitization of objects" [Kerouanton & Laroche, 2017]. The LabInVirtuo project focuses on human activity, the elicitation and restitution of expert knowledge in an EVIRS. It complements current approaches focused on the technical object, as in the issues of the Digital Heritage Reference Model (DHRM) metamodel [Laroche, 2017]. The interdisciplinarity of the project places it at the interface of several fields: history, anthropology and heritage of industries, VR/AR, including a sensory dimension that constitutes a field little explored [Simmonot et al., 2014] [Erdoen, 2017] [Bekele, 2018] yet necessary for a better virtual restitution of uses and knowledge incorporated, especially given the information that this dimension carries.
Making knowledge explicit in a virtual environment is the purpose of EVI [Aylett, 2001]. Some vr work [Bouville et al. 2015, Lourdeaux et al., 2017] is partly close to the goal of designing the EVI using a metamodel. These templates only allow the end user to access knowledge but not to create it in real time. For this we propose to use autonomous Conversational Agents (ACA) as an interface in the EV between the user and the digital historical corpus. ACAs appear more and more in THE EVI by taking the role of companion or tutor or psychologist listening [Gratch et al., 2014] by bringing undeniable benefits
[Rowe et al., 2007] [Kokane et al., 2014]. For the interaction to be effective and satisfactory, an ACA must be able to speak but also to listen, modify and adapt its behavior according to that shown by the user [Gratch et al., 2006]. However, the behaviors of agents developed in this area focus on the feedback exchanged between the participants during the dialogue but rarely on the semantics of the content of the messages exchanges or in the knowledge base of the agent. That is what we want to achieve.
Objective of the thesis: The objective of the thesis is to design the tools and think about the interactions that will make it possible to make the bidirectional link between the digital corpus and the virtual environment. The digital corpus corresponds not only to historical data and sources, but also to the technical devices for storing and accessing them. We postulate here that EV is a more natural set [interface, interactions] (compared to conventional tools) to navigate, simulate and augment this digital corpus. The thesis will contribute to the tasks identified in the ANR project.
The first task is to create the methods to make the bidirectional link between the digital corpus and the virtual environment by proposing multisensory interactions. It is a question of generating the informed virtual environment from the data of the digital corpus but also of proposing the metaphors to interact in the EV. We first consider (1st level) 3D, sound, gesture, haptic, tangible and, as far as possible, smells and tastes. For this we will rely on the ontologies developed by the project partners. We are then interested (2nd level) in the management of human activities within the environment. An activity here is for example the arrangement of actions of operators whose objective is to manipulate the technical system.
This level relies heavily on the first level because an operation can change the geometry (position, orientation) of a part of the system, generate a sound and require the realization of a gesture by an operator (played by an autonomous agent or by a user of the LabInVirtuo). Finally, the last level concerns the management, in the informed virtual environment, of the quality and integrity of the information contained in the digital corpus. For example, we will use visual metaphors to represent the degree of confidence of information (more or less transparent object for example). The most original point of this task will be the ability to enrich and modify the digital corpus from the virtual environment. For the first level (3D objects, sounds...), it is not necessarily relevant to modify the raw data within the EV on the other hand it becomes relevant for semantic information. Conversely, it is more relevant to describe an activity, operations and gestures in the virtual environment by demonstration.
The second task is to design interactions with the Animated Chatbot (ACA) and EV. The objective of the ACA is to act as an interface between users and knowledge in the virtual environment that is informed, originating or to be added to the digital corpus. The first objective is to provide the ACA with the ability to understand the questions or information provided by the user and to verbalize the answers or information of the ACA. The verbalizations of the ACA will be based on the techniques of classic TextToSpeech but the content will come from the reasoning of the ACA on the knowledge of the informed virtual environment. The goal here is not to work on natural language processing. We will build on results already proven by the consortium partners or within the scientific community.
The originality here is that we can imagine a grammar based on the metamodel used to design the informed virtual environment. The second objective of this task is to design the ACA reasoning algorithm based on knowledge and user interactions within the EV. The principle is to be able to propose new modes of interaction (haptic, gestural, tangible, ...) to question or inform the knowledge base via the agent. The agent's reasoning should also be used to generate new interactions between the ACA and users. For example, the ACA should be able to answer users' questions, identify gaps in the knowledge base (a 3D object without semantic information) or inconsistencies (an object named differently by several users). From these new interactions will emerge new knowledge that will then also be transmitted to the digital corpus.
Several types of distributed situations will be created based on the use cases planned by the partners, but we imagine situations of collaboration between an on-site user with an AR device and another in a CAVE-type device or situations with several users wearing VR headsets or a single user viewing the scene and interacting using an interactive table with tangible objects.
Expected skills and required degree: This thesis is open to all candidates with a master's degree or engineers in the fields of computer science.
Good knowledge of algorithmics and computer development is mandatory.
Experience in the field of Virtual and/or Augmented Reality will be appreciated.
A first use of tools such as Unity3D or Unreal will also be relevant.
As the project is by nature at the interface between computer science and the social sciences, an appetite for the fields of history, archaeology, heritage and mediation is also important.
Date and Place of the thesis: The thesis is scheduled over 3 years from February 1, 2022.
The thesis will largely take place at the Lab-STICC ENIB in Brest (see https://www.brestlife. fr/).
More specifically, the work will take place within the CERV (European Virtual Reality Centre) technological platform.
However, trips of several weeks will be scheduled to the partners (LS2N Nantes, FEMTO Belfort, Musée des arts et métiers Paris).
Regular meetings with the directors (Brest/Nantes) and supervisors (IT/Digital Humanities) will take place.
Application: Please send your application to QUERREC Ronan [email protected], LAROCHE Florent florent.laroche@ls2n. fr
Two PhD projects are proposed by the LILaC research group (www.irit.fr/LILaC/site/) at IRIT, at the University of Toulouse, France:
1) Belief revision for modal and epistemic logic. Supervisory team: Johnathan Ben-naim and Andreas Herzig ([email protected] , [email protected])
2) Model checking and automated mechanism design. Supervisory team: Guillaume Feuillade and Laurent Perrussel ([email protected], [email protected])
The details of the proposed projects are in attachment. The selection of applicants will be in two steps: first at department level (April), then an interview at the doctoral school level (May). For more information and for applications please write to the contact persons of the relevant project.
Umberto Grandi IRIT - Université Toulouse 1 Capitole Email: [email protected]
www.irit.fr/~Umberto.Grandi 0033 (0) 5 67 73 29 61