PhD Topics

PhD topics

DC1 - Situation Awareness and Complex Scene Understanding (WP3)

DC1 – This research stands pivotal in bridging technology and domestic safety, crafting a system capable of proactive risk management, thereby reducing domestic incidents and enhancing the quality of life for occupants. Especially relevant for those considered more vulnerable (such as the elderly and children), this research not only contributes to individual household safety but also brings about advancements in assistive robotics, broadening application horizons. This research posits the integration of robust robotic systems, leveraging advanced technologies like Metric-Semantic SLAM and Direct Situational Comprehension (DSC), as a pivotal strategy to not only identify but also proactively manage potential domestic risks.

Host Institution: Università degli Studi di Napoli Federico II

PI: Prof. Daniel Riccio

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Motion planning for scene understanding (United Robotics Group GmbH – 12m); Evaluation in realistic settings (Alisys Digital S.L.U. – 6m).

Essential Skills Requirements:
– Computer Science/Engineering degree
– Excellent computer programming skills

Desirable:
– Computer Vision
– Experience in Human-Robot Interaction research methodologies (e.g. running user studies, working in multidisciplinary teams)
– Practical experience in AI, Machine Learning or Computer Vision

DC2 - Tracking and analysis of temporal properties of human behaviour and non-verbal signals (WP3)

DC2 – The objective of this project is to develop a robotic system that is able to understand and appropriately interpret human multimodal cues for predicting the intentions and future actions of other actors while sharing the same environment with the robot. The project intends to investigate how multi modalities can leverage natural mechanisms to understand users’ intentions, beliefs and understanding of the situational context. To this extent, this research project will effectively mix MLLM and Transformer models with cross-modal attention layers for simultaneously identifying and interpreting human social verbal and non-verbal cues, understanding people’s intentions, and endowing robots with natural intelligence so that they can select their subsequent actions.

Host Institution: Università degli Studi di Napoli Federico II

PI: Prof. Silvia Rossi

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Working on integrating human behaviour recognition in social navigation (NAVER France – 12m); Sensing beyond vision (RT Corporation – 6m).

Essential Skills Requirements:
– Computer Science/Engineering degree
– Excellent computer programming skills

Desirable:
– Experience in Human-Robot Interaction research methodologies (e.g. running user studies, working in multidisciplinary teams)
– Practical experience in AI, Machine Learning or Computer Vision

DC3 - Adaptation of Social Cues (WP3)

DC3 – A system that can reliably generate social behaviours in adaptation to the social behaviour of a person. The system should exhibit the robustness, generalisability and utility that is expected of a commercially available robot platform with real end-users in the wild. The work is validated with user studies on the effectiveness of the adaptive behaviour on relevant interaction parameters, such as trust or likability. Additionally, the project is expected to live up to criteria for ethical and cultural sensitivity, not just technical advancements, by following guidelines for responsible design and deployment of socially intelligent robots as well as design principles to foster trust and acceptance in diverse contexts.

Host Institution: Furhat Robotics AB

PI: Prof. Gabriel Skanze

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Integration of non-verbal cues in explanation generation (Universität Hamburg – 12m); Social cues and dialogue with norms (École nationale supérieure de techniques avancées Paris – 6m).

Essential Skills Requirements:
– MSc in computer science, cognitive science, or closely related field
– Experience of machine learning and AI in practical applications

Desirable:
(note that we do not expect students to comply with every of these desirable skills; they are merely a list of extra skills that would be especially welcome for this project)
– Experience in human-robot interaction
– Experience in human factors/UX

DC4 - Social Embeddings (WP3)

DC4 – This project targets two main research objectives: (1) how to build compact, yet semantics-preserving, embeddings to represent arbitrary social environments; how to fully characterize these embeddings, including their latent semantics; (2) precisely frame the socio-cognitive skill of social awareness enabled by social embeddings, and demonstrate it on social robots. The DC will particularly focus on social dynamics, by characterizing the trajectories of on-going social situations in the embedding space; discontinuities in the embedding space, that might represent unexpected changes of social dynamics; and social situation predictions, by extrapolating trajectories in the embedding space.

Host Institution: PAL Robotics SL

PI: Dr. Severin Lemaignan

PhD Enrolment: Universitat Politècnica de Catalunya

Planned secondment(s): Social embeddings and RL (Agencia Estatal Consejo Superior de Investigaciones Científicas – 12m); Social perception beyond vision (University College Cork – 6m).

Essential Skills Requirements:
– MSc in artificial intelligence, machine learning or closely related field
– Demonstrable evidence of experience with at least one machine learning framework (eg pytorch, keras, etc)
– Expertise in Python and/or C++ programming

DC5 - Securing multimodal service robots to protect user’s privacy (WP3)

DC5 – The research aims to develop an innovative security and privacy framework for social and care robots, particularly in healthcare settings. Key objectives include preserving user privacy, designing user-centric interactions, and creating seamless and continuous authentication methods. The research emphasizes the use of adaptive biometric data, such as gait patterns, collected by the robot itself. Efficient data access control for unstructured data, considering diverse stakeholder needs, is also a focus. Scalability and the prevention of unintended data disclosure are essential goals. Additionally, the research seeks to enhance robots’ understanding of cultural sensitivity and social behaviours to build trust and acceptance among users.

Host Institution: Sheffield Hallam University

PI: Prof. Alessandro Di Nuovo

PhD Enrolment: Sheffield Hallam University

Planned secondment(s): Collaborative intelligence for H-R cooperation (Honda Research Institute Europe GmbH – 12 m); Context privacy in HRI interaction in open spaces (PAL Robotics SL – 6m).

Essential Skills Requirements:
– First-class honours degree in Computer Science/Engineering or equivalent
– Excellent computer programming skills
– Proficiency in English, with an overall IELTS score of 7.0 or above, and at least 6.5 in each component, or an accepted equivalent certificate

Desirable:
– Strong master’s degree (or expectation of the same) or equivalent
– Experience in programming robotic platforms
– Experience in applying research methodologies, particularly those required for Human-Robot Interaction (e.g., running user studies, working in multidisciplinary teams)
– Practical experience in programming and executing real-world applications of AI, Machine Learning, and/or Computer Vision
– Experience in teamwork within interdisciplinary projects

DC6 - Robot Behaviour with Social Norms Competences: Generation and Adaptation (WP4)

DC6 – Developing socially aware robots requires the ability to interpret context, follow social norms, recognize cues, and process feedback from both the environment and human interactions. Effective communication through verbal and non-verbal behaviors is also essential. While some research addresses social norms, few robots exhibit comprehensive social awareness.
This thesis explores how robots can learn and adapt to social norms—collective behaviors shaped by approval and disapproval—through reinforcement learning and game theory. It will examine both widely accepted conventional norms and essential norms that structure interactions. The key objectives are to:
(i) Define a model of social norms;
(ii) Develop a method based on reinforcement learning and game theory to enable the robot to learn and adapt its behaviour based on the context and social norms, and
(ii) Evaluate the model’s performance in real-time situations, thereby enhancing the competence and adaptability of robotic systems with a focus on adhering to social norms.

Host Institution: École nationale supérieure de techniques avancées Paris

PI: Prof. Adriana Tapus

PhD Enrolment: Institut Polytechnique de Paris

Planned secondment(s): Social Norms and privacy (NAVER France – 12m); Social awareness in service applications (RT Corporation – 6 months).

Essential Skills Requirements:
– MSc degree with a strong background in one or multiple of the following topics: machine learning, human-robot interaction (social interaction, …), human modeling, social psychology and social norms, personalized multimodal interaction, and machine learning
– Strong skills in C++ and Python

Desirable:
– Experience with ROS

DC7 - Navigation Styles for Guiding Robots (WP4)

DC7 – This research is set to explore the array of guiding tasks and the corresponding navigation styles that a robot should adopt to facilitate these tasks effectively. The study will analyse how different navigation styles can serve as non-verbal communication cues, potentially influencing the type of interaction between the robot and the person being guided. The aim is to understand and implement navigation behaviours that not only guide but also signal the robot’s intentions and encourage appropriate human responses.

Host Institution: NAVER France

PI: Dr. Mattia Racca

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Communication and legibility of guiding intent (Università degli Studi di Napoli Federico II – 12m); Evaluation of social intelligence in HRI (Korea Institute of Science and Technology – 6m).

Essential Skills Requirements:
– Robotics degree (or Computer Science/Engineering degree with courses on robotics)

Desirable:
– Research experience with Robot Navigation and Planning (e.g. a master thesis on such topic)
– Experience in Human-Robot Interaction research methodologies (e.g. running user studies, working in multidisciplinary teams)
– Practical experience in Machine Learning or Computer Vision

DC8 - Adaptive and adaptable robot decision making for service robots (WP4)

DC8 – To propose a new framework to learn and combine the preferences of user and bystanders in a service scenario. An offline phase will be used to bootstrap the preferences and compute the initial policies and the objective function using an Inverse Reinforcement Learning approach. In the online phase, the system will use the new experiences to refine the models. The framework will include an elicitation system to start interactions and eventually actively ask for missing information about the potentially incompatible preferences of the user.

Host Institution: Agencia Estatal Consejo Superior de Investigaciones Científicas

PI: Dr. Guillem Alenyà

PhD Enrolment: Universitat Politècnica de Catalunya

Planned secondment(s): Social embeddings and RL (PAL Robotics SL – 12m); Multi-robot control for preference elicitation (Bettering Our Worlds – 6m)

Essential Skills Requirements:
– Msc in Computer Science, Computer Engineering, Applied Mathematics, Artificial Intelligence, Machine Learning or a related field
– expertise in at least one of the following programming languages: python or C++
– knowledge of ROS
– good written and spoken English

Desirable:
– Experience in Machine Learning / Reinforcement Learning
– Experience in Experimental Design (designing, conducting, and analyzing human or robot-in-the-loop experiments)
– Experience in Human-Robot interaction

DC9 - Social awareness as a form of internal simulation (WP4)

DC9 – Social awareness involves understanding the environment and its inhabitants using the robot’s internal models. This research aims to unveil the potential of inner simulators embedded in a robotics cognitive architecture as the key facilitators of social awareness. This endeavour will explore the links between Bayesian and Active Inference and differentiable internal simulators within an advanced distributed cognitive architecture. Using this conceptual framework and the CORTEX architecture, the project will pursue the hypothesis that social awareness is a form of prediction-updating dynamics implicit in the architecture that continually aligns the robot with its environment. As a consequence of this self-maintaining loop, the robot will generate epistemic actions to gather information about the perceived objects and people, and transform discriminative bottom-up percepts into predictive stable beliefs in its internal model. Models of humans will be extended into simple intentional engines and added to the inner simulator to predict a small range of human actions in specific contexts. These new capabilities of the CORTEX architecture will constitute a firm step towards socially aware robots

Host Institution: Universidad de Extremadura

PI: Prof. Pablo Bustos Garcia de Castro

PhD Enrolment: Universidad de Extremadura

Planned secondment(s): Inner simulation and social aspects (PAL Robotics SL – 12m); Simulation for social navigation (NAVER France – 6m).

Essential Skills Requirements:
– MSc in robotics, artificial intelligence or a closely related field
– Excellent computer programming and mathematical skills

Desirable:
– Experience in programming robotic platforms
– Practical experience in Robotics, AI or Computer Vision
– Experience in teamwork within interdisciplinary projects

DC10 - Intention Recognition and ToM for Proactive Behaviour (WP4)

DC10 -The objective of this research is to use the ability to form a Theory of Mind and recognise others’ intentions, actions and behaviours for enabling robots to be proactive and correctly plan their actions in a socially acceptable way. This research is fundamental not only for planning current and future actions of a robot, but also for ensuring that these actions and behaviours are not intrusive. This research project develops new approaches that bridge together Knowledge Representation and Machine Learning techniques with effective solutions for fast and accurate human intent recognition of users and bystanders integrating symbolic and knowledge-driven AI and sub-symbolic and data-driven AI.

Host Institution: Università degli Studi di Napoli Federico II

PI: Prof. Silvia Rossi

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Proactive behaviour in HRI and Dialogue (Furhat Robotics AB – 12m); ToM in non-cooperative tasks (Honda Research Institute Europe GmbH – 6m).

Essential Skills Requirements:
– Computer Science/Engineering degree (also students from psychology and cognitive science could be considered)
– Excellent computer programming skills

DC11 - Decision Making to Leverage Artificial Trust (WP5)

DC11 – The research aims to advance Artificial Trust (AT) in Human-Robot Interaction by integrating Theory of Mind (ToM) and enhancing trust assessment models. By combining observable behaviours, ToM insights, and internal trust factors, the goal is to create a comprehensive trust assessment framework for robots. This approach allows robots to fine-tune their responses, considering human needs and preferences. The research also seeks to develop a novel artificial cognitive architecture informed by machine learning, psychology, and social sciences. Experimental testing with a humanoid social robot will evaluate trustworthiness assessment accuracy and its impact on task performance.

Host Institution: Sheffield Hallam University

PI: Prof. Alessandro Di Nuovo

PhD Enrolment: Sheffield Hallam University

Planned secondment(s): Robot-agnostic AT architecture in simulation (Bettering Our Worlds – 12 m); Collaborative intelligence in shared environments (Honda Research Institute Europe GmbH – 6m).

DC12 - Explaining and learning new grounded robot knowledge (WP5)

DC12 – To investigate how a robot should explain acquired knowledge, as well as behavioural change, to its user. Reasoning in robotic systems relies on specific, e.g. compositional and symbolic, representations of objects, space, tasks, actions and the agent. Since post-hoc reasoning can change a large language model’s (LLM) decision, a recursive in-depth reasoning process based on chain-of-thought prompting will be performed before any decision is accepted. The reasoning trace will be stored in short-term memory to reconcile with any required post-hoc reasoning. In case of conflict, any additionally considered facts such as externally accessed knowledge, as well as differences in belief and decisions made, will be communicated to the user.

Host Institution: Universität Hamburg

PI: Prof. Stefan Wermter

PhD Enrolment: Universität Hamburg

Planned secondment(s): Nonverbal behaviour for explanations (Furhat Robotics AB – 12m); Explanations in non-interactive tasks with bystanders (United Robotics Group GmbH – 6 months)

Essential Skills Requirements:
– MSc or demonstrated equivalent in Artificial Intelligence, Computer Science or Engineering with a focus on Intelligent Systems, Intelligent Robotics, Natural Language Processing or Neural Networks
– Excellent programming skills (Python, C++, PyTorch/Tensorflow, Machine Learning Frameworks, ROS etc.)
– Expertise in at least one of neural networks or intelligent robotics

Desirable:
– Experience with LLM prompting via API is a plus
– Existing publications

DC13 - Contextual Dependency of Privacy (WP5)

DC13 – The research will investigate the factors influencing office privacy attitudes to inform the creation of a versatile privacy toolkit. This toolkit, grounded in a socio-technical study, will be designed to understand and implement privacy in office environments and adaptable for use in other contexts, such as hospitals. It will focus on identifying key privacy elements that affect the integration of social robots in professional settings. This set of practical tools will enhance the ability of social robots to respect and maintain privacy constraints in different environments.

Host Institution: NAVER France

PI: Dr. Maria Antonietta Grasso

PhD Enrolment: Università degli Studi di Napoli Federico II

Planned secondment(s): Privacy settings and authentication (Sheffield Hallam University – 12m); Context recognition for privacy assessment (Università degli Studi di Napoli Federico II – 6m)

Essential Skills Requirements:
– Computer science/Engineering degree

Desirable:
– Research experience with robot navigation and privacy considerations in AI or robotics (e.g. a master thesis on such topic)
– Practical experience in Machine Learning
– Familiarity with user studies and interdisciplinary research

DC14 - Security and safety for trustworthy robots (WP5)

DC14 – DC14 will explore the relationship between safety and security using operational modes. The Operational Modes are useful to establish access to certain sensitive privacy information and the tasks that a robot can perform based on the robot’s level of security. During the operation of a robot, if it considers that its safety may have been compromised, it can establish operating modes in which the camera is not used or that the robot cannot access certain areas of the environment until the appropriate security recovery level is reached. Methods form the general XAI field will be investigated as tools for analysing the behaviour of intelligent systems.

Host Institution: Universidad de León

PI: Prof. Vicente Matellán

PhD Enrolment: Universidad de León

Planned secondment(s): Safety in IoT environments with robots (Alisys Digital S.L.U. – 12m); Trustworthiness in social navigation (NAVER France – 6m).

Essential Skills Requirements:
– Computer programming experience.
– Knowledge of ad hoc vision systems for autonomous robots.
– Team working abilities.

Desirable:
– Curiosity
– Spanish basic communication skills (at least an A1 level)
– Basic knowledge of ROS
– Basic knowledge of behavior trees

DC15 - Awareness as a Platform Independent Tool (WP5)

DC15 – The goal of this project is to investigate and develop a software infrastructure that enables BOW-compatible applications to run seamlessly across multiple robot types, thereby fostering a more universal approach to robotics software development. We plan to enhance collaborative research efforts by investigating a cloud-native, low-latency communication framework, allowing for shared access to remote robots between research groups. Lastly, we will research and develop solutions for the multi-entity control of robots crucial to the deployment of multiple software systems competing for the same hardware resource.

Host Institution: Sheffield Hallam University

PI: Prof. Alessandro Di Nuovo

PhD Enrolment: Sheffield Hallam University

Planned secondment(s): Development of BOW-compatible applications (Bettering Our World – 18 m); Inner simulations for multiple robots (Universidad de Extremadura – 6 months); Remote control and security (Universidad de León – 6 months).