Titulo Estágio
Evaluating Human-Robot Collaboration through Virtual Reality Scenarios
Local do Estágio
DEI-FCTUC
Enquadramento
This dissertation will be developed in the context of the project NEUROCOBOTS - Developing brain-machine interface for dynamic human-robot collaboration - which seeks to explore a novel way for efficient and adaptive human-robot collaboration.
Naturalistic feeling is a problematic factor limiting human-robot interaction. For example, one could expect that as robots become more human-like, the quality of human interactions with them would steadily improve. However, this is not always true – if humanoid robots resemble humans too close, human observers perceive them as strange and unpleasant to interact with. This effect is commonly designated as “uncanny valley”. While the “uncanny valley” effect has been described for social interaction between humans and robots, when in virtual scenarios, nothing is known about its impact on human motor performance.
Likewise, although it was previously reported that humans operating assistive robots perform better if these robots follow human-like movement patterns (e.g. the relationship between curvature and speed of movements), it is not known whether the same applies to scenarios where humans and cobots work autonomously (just like while cooperating). Human actions are predictable in the sense that arm joint configurations define possible movement degrees of freedom, allowing the brain to construct predictive models of an ongoing action of the other person based on natural human motor repertoire. For example, the human hand usually follows a smooth, predictable path, even if complex arm joint configurations are needed to follow the path. This makes hand-path prediction easier for the brain as the hand is usually most important for understanding an action goal. For observing robot actions, this is less obvious, as robotic arms do not have the default biomechanical design constraints the human arm has, and can execute much more complex movements. Yet, the correct prediction of other agent’s movements is needed for adapting one’s own actions and, as such, efficient cooperation. That’s why it is of vital importance to understand whether the human brain is capable of perceiving robotic actions with the same proficiency, as it can perceive actions of other humans and how this action perception is influenced by different robot designs.
Objetivo
The goal of this dissertation is to evaluate the usability and user experience of different VR scenes to study the “uncanny valley” effect and to determine whether there is a learning effect for non-human-like actions.
The project will provide readily developed VR scenarios, displaying human movement parameters using a hand motion capture system and eye tracking. The student will develop a suitable usability and user experience testing protocol. This protocol will then be applied to evaluate and compare the VR Scenarios. The student will analyse motion capture and eyetracking data and, optionally, neural and biosensor signals to complete the behavioral data. The goal is also to compare human self-reports (e.g. feeling of comfortable cooperation) with their actual hand and eye movements for different characteristics of virtual cobots in search for the “uncanny valley” effect in motor behavior. The presence of a learning effect for non-human-like actions (i.e. if human performance gets better over time in trials with non-human-like cobot movements) will also be assessed.
Plano de Trabalhos - Semestre 1
- Review literature on collaborative robotics and the “uncanny valley” effect.
- Review literature on usability and user experience test experiments for Virtual Reality and Collaborative robotics.
- Review and analyse the NEUROCOBOTS project VR scenarios.
- Design a usability and user experience testing protocol.
- Intermediate dissertation document writing.
Plano de Trabalhos - Semestre 2
- Apply the usability and user experience testing protocol and perform a series of comparative experiments NEUROCOBOTS project VR scenarios.
- Analyse, document and extract results from data collected.
- Extract and discuss conclusions and draw preliminary recommendations on human-robot collaboration scenarios.
- Dissertation writing.
Condições
The project would be performed under constant support from the supervisors and in collaboration with other team members. The student should have or be willing to learn relevant analytical skills (such as motion capture data analysis) as well as be able to work and communicate with study participants and other team members. Strong interest in the field of human-machine interactions and robotics is desirable. English proficiency is required.
Observações
This dissertation will be advised by Paula Alexandra Silva and co-advised by Artur Pilacinski. For more information email us on paulasilva@dei.uc.pt and art.pilacinski@gmail.com
The project is jointly developed by University of Coimbra, Institute for Systems and Robotics and University of Madeira, together with international partners.
Orientador
Paula Alexandra Silva
paulasilva@dei.uc.pt 📩