Human-robot interaction requires building a joint understanding of context, facilitating collaboration naturally and seamlessly on tasks, e.g. by joint goal setting, communicating progress or clarifying the user’s intention. To achieve the ability of natural command, control, and feedback in real world scenarios requires the construction of user interaction models supported by spatial modelling and reasoning, that can link a detailed digital landscape to real world concepts.
Stipend and International Fees, subject to availability of funding
There is only one studentship available for stipend and international fees but a number of projects that are eligible for this funding. In the event that this studentship is allocated to a project, funding for stipend and home fees only will be available for the remaining projects.
Critical illness can affect individuals at any age and for a wide range of medical and surgical conditions. Recovery can be prolonged, and complicated by fatigue, impaired attention and limited engagement with rehabilitation for physical and mental health reasons. Socially assistive robots provide an opportunity for bespoke rehabilitation programmes to be developed by health care professionals, then delivered by the robot, from the time of recovery from critical care, through the rest of the inpatient journey, to the transition home.
Interactions with current AI agents (e.g., Amazon Alexa, Google Home, Apple Siri) are limited to single-turn simple tasks such as asking about the weather, listening to a song, or telling a joke. What they are currently lacking is a more in-depth multi-turn conversation on wider domains (e.g., talking about the news) that entail follow-up questions, retrieving information from knowledge bases (e.g., WikiData), texts (e.g., news articles) and performing common-sense reasoning (e.g., if-then clauses).
In this project, we will explore a Neurorobotics approach to develop robotic controllers to scenarios which behavioural responses must be fast and precise whilst limited to strict energy constraints. This will be accomplished by the use of Evolving Spiking Neural Networks (ESNN) models.
Results from this project would be applicable to robotics in search and rescue missions, dangerous scenarios, critical operations and a plethora of human-robot-interaction scenarios, where enhanced autonomy, robustness and rapid responses are of vital importance.
Mobile robots are used in a variety of applications inside and outside of constrained areas. When these robots move to the real-world, they will have to navigate around people in a manner that is not only safe from a technical point of view but also makes the person feel safe. This feeling of safety is paramount for the comfort the person experiences when interacting with the robot in this way.
There is an increasing demand for both robust and explainable deep learning systems in real world applications. Recently, it has been shown that malicious examples can be crafted to fool deep learning models, the so-called adversarial attacks. For example, one can fool a self-driving car to drive over the speed limit by making minor modifications on the speed limit sign. Another extreme example is the one-pixel attack: an adversarial attack can be made by altering only one pixel of an image in order to fool deep neural networks.
As machine learning algorithms find their ways in safety-critical systems, such as autonomous cars, robot nurses, conversational agents, the question of ensuring their safety and security becomes important. At the same time, neural networks are known to be vulnerable to adversarial attacks --- a special kind of crafted inputs that cause unintended behaviour in trained neural networks. Due to these two factors, neural network verification has become a hot topic in both machine learning and verification communities.
Social Robots and other artificial agents are often designed to have a digital persona. A digital persona can be viewed as a composite of elements of identity (such as demographics and background facts), behaviour, and interaction style.
These are deliberate design choices, often influenced by common perceptions, such as female voices being more `pleasant' when choosing a synthesised voice and gender.However, these design choices bear the risk of reinforcing social stereotypes, according to the UNESCO (West et al., 2019).
Robotic applications spread to a variety of application domains, from autonomous cars and drones to domestic robots and personal devices. Each application domain comes with a rich set of requirements such as legal policies, safety and security standards, company values, or simply public perception. They must be realised as verifiable properties of software and hardware. Consider the following policy: a self-driving car must never break the highway code.
Safety critical robotic and autonomous systems, such as Unmanned Air Vehicles (UAVs) that operate beyond visual line of sight, require the highest level of certification. Certifiers are concerned with how such systems behave within their environment – as defined by system wide requirements, e.g. compliance with the rules-of-the-air (i.e. SERA). In contrast, software developer’s focus on specifications - how the system software should behave based upon operational modes and input signals. Many catastrophic system failures, e.g.