Below you’ll find a list of some of our outputs- but do check back soon or follow us on X for our upcoming work, including our ARET study.

Specification

• AERoS: Assurance of Emergent Behaviour in Autonomous Robotic Swarms
• Link: https://link.springer.com/chapter/10.1007/978-3-031-40953-0_28
• If you’re interested in finding out more about: processes for the safety assurance of emergent behaviour in autonomous robotic swarms.
• Abstract: The behaviours of a swarm are not explicitly engineered. Instead, they are an emergent consequence of the interactions of individual agents with each other and their environment. This emergent functionality poses a challenge to safety assurance. The main contribution of this paper is a process for the safety assurance of emergent behaviour in autonomous robotic swarms called AERoS, following the guidance on the Assurance of Machine Learning for use in Autonomous Systems (AMLAS). We explore our proposed process using a case study centred on a robot swarm operating a public cloakroom

 

• Emergence of norms in interactions with complex rewards
• Link: https://link.springer.com/article/10.1007/s10458-022-09585-3
• If you’re interested in finding out more about: novel agent-based modelling approaches to investigating norm emergence.
• Abstract: Autonomous agents are becoming increasingly ubiquitous and are playing an increasing role in wide range of safety-critical systems, such as driverless cars, exploration robots and unmanned aerial vehicles. These agents operate in highly dynamic and heterogeneous environments, resulting in complex behaviour and interactions. Therefore, the need arises to model and understand more complex and nuanced agent interactions than have previously been studied. In this paper, we propose a novel agent-based modelling approach to investigating norm emergence, in which such interactions can be investigated. To this end, while there may be an ideal set of optimally compatible actions there are also combinations that have positive rewards and are also compatible. Our approach provides a step towards identifying the conditions under which globally compatible norms are likely to emerge in the context of complex rewards. Our model is illustrated using the motivating example of self-driving cars, and we present the scenario of an autonomous vehicle performing a left-turn at a T-intersection

 

• On Specifying for Trustworthiness
• Link: https://arxiv.org/abs/2206.11421
• If you’re interested in finding out more about: the key specification challenges in the domains of resilience, trust, functionality, verifiability, security, and governance and regulation of Autonomous Systems.
• Abstract: As autonomous systems (AS) increasingly become part of our daily lives, ensuring their trustworthiness is crucial. In order to demonstrate the trustworthiness of an AS, we first need to specify what is required for an AS to be considered trustworthy. This roadmap paper identifies key challenges for specifying for trustworthiness in AS, as identified during the “Specifying for Trustworthiness” workshop held as part of the UK Research and Innovation (UKRI) Trustworthy Autonomous Systems (TAS) programme. We look across a range of AS domains with consideration of the resilience, trust, functionality, verifiability, security, and governance and regulation of AS and identify some of the key specification challenges in these domains. We then highlight the intellectual challenges that are involved with specifying for trustworthiness in AS that cut across domains and are exacerbated by the inherent uncertainty involved with the environments in which AS need to operate

 

Ethics

• Hasta la vista baby: why we should dispense of “autonomy” in “autonomous systems”
• Link: https://research-information.bris.ac.uk/en/publications/hasta-la-vista-baby-why-we-should-dispense-of-autonomy-in-autonom
• Summary: The term ‘autonomous’ is used in tech circles to describe the capabilities of systems. In this paper, we argue that, in so doing, the term ‘autonomy’ is misused.

 

• Clinicians and AI use: where is the professional guidance?
• Link: https://jme.bmj.com/content/early/2023/08/22/jme-2022-108831
• If you’re interested in finding out more about: the clinical, professional and reputational safety risks associated with deficits in professional guidance for clinical users of AI clinical decision support systems.
• Abstract: With the introduction of artificial intelligence (AI) to healthcare, there is also a need for professional guidance to support its use. New (2022) reports from National Health Service AI Lab & Health Education England focus on healthcare workers’ understanding and confidence in AI clinical decision support systems (AI-CDDSs), and are concerned with developing trust in, and the trustworthiness of these systems. While they offer guidance to aid developers and purchasers of such systems, they offer little specific guidance for the clinical users who will be required to use them in patient care. This paper argues that clinical, professional and reputational safety will be risked if this deficit of professional guidance for clinical users of AI-CDDSs is not redressed. We argue it is not enough to develop training for clinical users without first establishing professional guidance regarding the rights and expectations of clinical users.

 

Regulation

• AI regulation: a pro-innovation approach – policy proposals: TASHub Response
• Link: https://eprints.soton.ac.uk/478329/
• Summary: Response to open consultation from: Department for Science, Innovation and Technology and Office for Artificial Intelligence

 

Verification

• On Determinism of Game Engines used for Simulation-based Autonomous Vehicle Verification
• Link: https://arxiv.org/abs/2104.06262
• If you’re interested in finding out more about: the potential causes of non-deterministic behaviours in game engines.
• Abstract: Game engines are increasingly used as simulation platforms by the autonomous vehicle (AV) community to develop vehicle control systems and test environments. A key requirement for simulation-based development and verification is determinism, since a deterministic process will always produce the same output given the same initial conditions and event history. Thus, in a deterministic simulation environment, tests are rendered repeatable and yield simulation results that are trustworthy and straightforward to debug. However, game engines are seldom deterministic. This paper reviews and identifies the potential causes of non-deterministic behaviours in game engines. A case study using CARLA, an open-source autonomous driving simulation environment powered by Unreal Engine, is presented to highlight its inherent shortcomings in providing sufficient precision in experimental results. Different configurations and utilisations of the software and hardware are explored to determine an operational domain where the simulation precision is sufficiently low i.e.\ variance between repeated executions becomes negligible for development and testing work. Finally, a method of a general nature is proposed, that can be used to find the domains of permissible variance in game engine simulations for any given system configuration

 

• Engineering Responsible And Explainable Models In Human-Agent Collectives
• Link: https://research-information.bris.ac.uk/en/publications/engineering-responsible-and-explainable-models-in-human-agent-col
• If you’re interested in finding out more about: novel engineering methodology based on formal model checking as a step towards providing evidence for the certification of responsible and explainable decision-making within human-agent collectives.
• Abstract: In human-agent collectives, humans and agents need to work collaboratively and agree on collective decisions. However, ensuring that agents responsibly make decisions is a complex task, especially when encountering dilemmas where the choices available to agents are not unambiguously preferred over another. Therefore, methodologies that allow the certification of such systems are urgently needed. In this paper, we propose a novel engineering methodology based on formal model checking as a step toward providing evidence for the certification of responsible and explainable decision making within human-agent collectives. Our approach, which is based on the MCMAS model checker, verifies the decision-making behavior against the logical formulae specified to guarantee safety and controllability, and address ethical concerns. We propose the use of counterexample traces and simulation results to provide a judgment and an explanation to the AI engineer as to the reasons actions may be refused or allowed. To demonstrate the practical feasibility of our approach, we evaluate it using the real-world problem of human-UAV (unmanned aerial vehicle) teaming in dynamic and uncertain environments.

 

Swarm robotics

• Building Trustworthiness by Minimizing the Sim-to-Real Gap in Fault Detection for Robot Swarms
• Link: https://research-information.bris.ac.uk/en/publications/building-trustworthiness-by-minimizing-the-sim-to-real-gap-in-fau
• If you’re interested in finding out more about: how well metrics can transfer from simulation to a real-world setting.
• Abstract: As robot swarm applications move to the real-world, ensuring the safety of such systems will be critical for trust and adoption. Fault detection is an essential component in systems which require a level of safety. Previous work has identified metrics with high discriminatory power between faulty and normal states of a robot in the swarm. The method for identifying such metrics has been implemented in simulation. Here, we implement metric extraction in a real-world environment and evaluate whether the extracted metrics can overcome the “sim-to-real gap” – in other words how well it transfers from simulation to a real-world setting

 

• A Data-Driven Method for Metric Extraction to Detect Faults in Robot Swarms
• Link: https://research-information.bris.ac.uk/en/publications/a-data-driven-method-for-metric-extraction-to-detect-faults-in-ro
• If you’re interested in finding out more about: self-detection of faults in robots, including methods for identifying and evaluating metrics that may be used in building a model for fault detection.
• Abstract: Robot swarms are increasingly deployed in real-world applications. Making swarms safe will be critical to improve adoption and trust. Fault detection is a useful component in systems which require a level of safety: a key element of which are metrics that allow us to differentiate between faulty and normal (non-faulty) robots – metrics which are measurable on-board the individual robots for self-detection of faults. In this paper, we develop a method for identifying and evaluating such metrics and discuss how these metrics may be used in building a model for fault detection. We demonstrate this method for real-time error detection in a realistic use-case: intralogistics using swarms. We show that we are able to identify metrics of large effect size for various faults, demonstrating the potency of metrics selected in this way with a simple fault detection model

 

• Stochastic behaviours for retrieval of storage items using simulated robot swarms
• Link: https://research-information.bris.ac.uk/ws/portalfiles/portal/345093205/Full_text_PDF_final_published_version_.pdf
• If you’re interested in finding out more about: robot swarm simulations, particularly applications for retrieval and delivery of items.
• Abstract : Robot swarms have the potential to be used as an out-of-the-box solution for storage and retrieval that is low cost, scalable to the needs of the task, and would require minimal set up and training for the users. Swarms are adaptable, robust and scalable with a relatively low computational cost which makes them appropriate for this purpose. This project simulated a robot swarm with simple sensors and stochastic movement, collecting boxes from storage to deliver them to the user. We show in simulation that stochastic strategies based on random walk and probabilistic sampling of local boxes could give rise to competitive solutions to retrieve boxes and deliver them unordered, or following a predetermined order, within a storage scenario. The performance of the task is drastically improved using an additional simple bias rule which uses compass measurements and does not reduce the minimalism of the control. It is shown that swarm technology could provide an out-of-the-box system for storage and retrieval using only information local to each robot and with distributed control.

 

Soft Robotics

• Recycled Materials for Trustworthy Soft Robots
• Link: https://ieeexplore.ieee.org/document/9762170
• If you’re interested in finding out more about: the recycling of soft robotic materials, reducing the amount of new material needed, lowering production costs, and reducing harmful waste.
• Abstract: The ethical sourcing and disposal of materials has become ubiquitous and necessary across society. However, this trend towards increased sustainability has yet to find parallels in robotics research. Within the field of soft robotics, large quantities of silicones, rubbers and other elastomers are used to construct the various soft bodies and actuators. As the field grows, this will have a large negative impact on the environment. In this work we propose the effective recycling of elastomeric materials, thereby reducing the amount of new material needed, lowering costs and reducing the amount of harmful waste. We present a non-chemical process of elastomer recycle-and-reuse, where elastomeric material from old and broken soft actuators is ground into granules ranging from ≤1 mm in diameter to 3 mm in diameter and used to create new soft actuators without loss of function. Characterisation tests show that although ultimate yield strain and stress reduce with the percentage of recycled material, the silicone composites exhibit very comparable elastic properties to the pristine silicone. This suggests an effective recycling pipeline where materials are recycled into different, lower-risk, applications at each iteration. We demonstrate the effectiveness of this process in a high strain recycled bladder actuator and a soft gripper.

 

• Towards a Soft Exosuit for Hypogravity Adaptation: Design and Control of Lightweight Bubble Artificial Muscles
• Link: https://ieeexplore.ieee.org/abstract/document/9762121
• If you’re interested in finding out more about: how soft robotic exosuits can help maintain physical fitness and health in space exploration.
• Abstract: Lower body soft exosuits have been shown to improve the capabilities of humans in a wide range of applications, from rehabilitation to worker enhancement. Their light weight and ability to be easily sewn into fabrics make them attractive for both terrestrial and space exploration applications. One unaddressed challenge in space exploration is the prevalence of low (hypo) gravity conditions, which can have a serious deleterious effect on the human body. To address this challenge we propose the hypogravity exosuit (or HEXsuit), which can help maintain the physical fitness and health of inter-planetary travellers. A core component of the HEXsuit is compliant, comfortable and efficient soft robotic artificial muscles. A recently proposed pneumatic actuator, the Bubble Artificial Muscle (BAM), is particularly suited for integration into hypogravity exosuits. In this work we explore the design and control of lightweight BAM actuators. Characterisation results show that a thin actuator is capable of high contraction, while a thicker actuator can be used for high load applications. Two control modes were implemented: displacement control and force control. Both controllers achieve low steady state error and show high accuracy. The displacement controller is also shown to be capable of maintaining the required displacement while actively changing external loads, a typical use case within the proposed hypogravity HEXsuit. Soft Robotics, Bubble Artificial Muscles, pneumatic, artificial muscle, exoskeleton, exosuit, space exploration

 

• A Contact-triggered Adaptive Soft Suction Cup
• Link: https://research-information.bris.ac.uk/en/publications/a-contact-triggered-adaptive-soft-suction-cup
• If you’re interested in finding out more about: a gripper that has the ability to adapt to objects with complex and irregular shapes.
• Abstract: Suction adhesion is widely used by natural organisms for gripping irregular objects (e.g., rocks), but their artificial counterparts show less adaptation in the same situation. In addition, they can require complex sensing and control systems to function. In this paper, we present a contact-triggered suction cup with the ability to adapt to objects with complex and irregular shapes. The gripper has two states to adhere and release the object and the transformation from release to adhesion is passively triggered by the contact force, making it an autonomous gripper and removing the need for complex driven system. Once the suction cup experiences a contact force above a set threshold, it will automatically capture the contacting object. Only the resetting transformation from adhesion to release is actuated by a vacuum pump. The maximal suction force up to 15.1 N is generated on the non-flat surface with the suction cup diameter of 30 mm. The performance of this gripper is demonstrated on a 7 DoF robot arm which successfully picked up a variety of irregular objects. We believe that this contact-triggered gripper provides a new solution for low cost, energy-effective and adaptive soft gripping.

 

Uncrewed Air Vehicles

• Wind tunnel testing of an avian-inspired morphing wing with distributed pressure sensing
• https://research-information.bris.ac.uk/en/publications/wind-tunnel-testing-of-an-avian-inspired-morphing-wing-with-distr
• If you’re interested in finding out more about: testing the benefits of combining bio-inspired flight technologies in a wind-tunnel model.
• Abstract: Small fixed wing uncrewed air vehicles (UAVs) are often required to fly at low speeds and high angles of attack, particularly when operating in urban environments. This study focuses on the potential of combining two bio-inspired flight technologies to improve maneuverability under these conditions. The outstanding flight agility of birds is believed to be enabled by the capability to sense the airflow over their wings and morph their wing surfaces accordingly. To test the benefits of combining these abilities a wind tunnel model able to perform an avian-inspired wing sweep motion incorporating two arrays of pressure sensors was developed. Aerodynamic load results highlight strong changes to the pitching moment produced by the change in wing sweep angle. This suggests that wing sweep can be an alternative or complementary mechanism for pitch attitude control, improving control authority at high angles of attack. On the other hand, pressure sensing data shows the ability of these sensors to detect the fine details of the onset of aerodynamic stall. The combination of these two novel technologies is suggested as a potential method to improve UAV pitch control when flying at low speeds, when the aircraft is most susceptible to environmental disturbances.

 

• Sim-to-Real Transfer for Fixed-Wing Uncrewed Aerial Vehicle
• Link: https://ieeexplore.ieee.org/document/9882333
• If you’re interested in finding out more about: how training of uncrewed air vehicles can mitigate the gap between simulated and real environments.
• Abstract: Deep reinforcement learning has great potential to automatically generate flight controllers for uncrewed aerial vehicles (UAVs), however these controllers often fail to perform as expected in real world environments due to differences between the simulation environment and reality. This letter experimentally investigated how this reality gap effect could be mitigated, focusing on fixed-wing UAV pitch control in wind tunnel tests. Three different training approaches were conducted: a baseline approach that used simple linear dynamics, a high-fidelity modeling approach, and a domain randomization approach. It was found that the base line controller was susceptible to the reality gap, while the other two approaches successfully transferred to real tests. To further examine the controllers’ capabilities to generalize, a variety of configuration changes were experimentally implemented on the UAV, such as increased inertia, extended elevator area, and aileron offset. While the high-fidelity controller failed to cope with these changes, the controller with domain randomization maintained its performance. These results highlight the importance of selecting appropriate sim-to-real transfer approaches and how domain randomization is applicable to fixed-wing UAV control with uncertainty in real environments.

 

 

 

 

 

 

In a new paper published in BMJ’s Journal of Medical Ethics, the TAS functionality nodes’ Jonathan Ives, John Downer and Helen Smith explore the increased use of AI in healthcare and medical settings, and the lack of professional guidance around it.

Although AI has great potential to help improve medical care and alleviate the burden on healthcare workers, the authors argue that as there is no precedent for when AI or AI-influenced medical workers make a mistake, regulation should be developed as a priority to outline the rights and expectations of those working closely with it.

There have recently been reports from National Health Service AI Lab & Health Education England which focus on healthcare workers’ understanding and confidence in AI clinical decision support systems, and are concerned with developing trust in, and the trustworthiness of these systems. However while they offer guidance to aid developers and purchasers of such systems, they offer little specific guidance for the clinical users who will be required to use them in patient care.

The clinicians who will have to decide whether or not to enact an AI’s recommendations are subject to the requirements of their professional regulatory bodies in a way that AIs (or AI developers) are not. This means that clinicians carry responsibility for not only their own actions, but also the effect of the AI that they use to inform their practice.

The paper argues that clinical, professional and reputational safety will be risked if this deficit of professional guidance for clinicians, and that this should be introduced urgently alongside the existing training for clinical users.

The authors end with a call to action for clinical regulators: to unite to draft guidance for users of AI in clinical decision-making that helps manage clinical, professional and reputational risks.

More information

Read the full paper:

http://dx.doi.org/10.1136/jme-2022-108831

Read a blog about the paper on BMJ

 

All authors are fully or part funded via the UKRI’s Trustworthy Autonomous Systems Node in Functionality under grant number EP/V026518/1.

Helen Smith is additionally supported by the Elizabeth Blackwell Institute, University of Bristol via the Wellcome Trust Institutional Strategic Support Fund.

Jonathan Ives is in part supported by the NIHR Biomedical Research Centre at University Hospitals Bristol and Weston NHS Foundation Trust and the University of Bristol. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health and Social Care.

 

Matimba Swana, our TAS PhD student, joined other early career researchers in a “reverse think tank” project last year, organised by the Public Engagement team at University of Bristol and co-led by Kilter Theatre. Designed to be a creative approach to the new emphasis on responsible innovation, as well as encouraging debate about what that means, the reverse think tank challenges the notion of the “expert”. Eleven 15-18 year olds gathered to learn about current research which they then used in a collaborative storytelling activity, based on a Disney strategy they employed to think of new ideas for films.

Although the researchers weren’t present, the young people saw videos of Matimba’s SWARM study research, to get an idea of where the research might lead. She is looking at the ethics and regulations of nanoswarms, a future nanotechnology that could be developed to treat cancer, to help shape the first in-person clinical studies. The young people were excited about the possibility of the technology being used but were concerned about negative consequences such as autonomy and unknown long-term effects. Their story involved a man named Ralph, living in Bristol, in his early 50s. He’s wondering whether to get nanoswarm treatment for cancer, asks for advice from his wife, who is very distrustful of the medical establishment, and is “anti-vaxx” and his friend Bob, who believes this is Ralph’s chance to be immortal. This also highlighted the possible inequalities that could arise from nanoswarm use, and the general distrust of medical advances.

Matimba found the think tank outcomes really useful and says “Something that resonated with me was we conduct research that impacts the future of young people and yet their voices are often not included in this area of research. Engaging young people as peer reviewers of my research has been enriching and a chance to understand their thoughts and feelings about the technologies we are investigating that could impact their future.”

You can read more about it here

If you’d like to participate in our SWARM study focus groups, please go here.