Starting date, duration: 10/2018, 36 months

Details: This project focuses on the control of nonlinear systems through a multitude of methods.  Specifically, it focuses on nonlinear resonance control. Many of the systems worked with have been oscillatory ones featuring nonlinear resonance behaviors such as Hopf Bifurcations and Saddle-Node Bifurcations which have bizarre behavior, making them notoriously difficult to suppress.

Current work involves optimising the positive family of feedback controllers, known as; PAVPF (Positive Acceleration Velocity Position Feedback) , PVPF (Positive Velocity Position Feedback) and PPF (Positive Position Feedback) on a linear second order system. 

Additional  work involves understanding this family of controllers on a nonlinear cantilever beam featuring a Saddle-Node Bifurcation.

Funder: School of Engineering, University of Aberdeen (PhD studentship)

Starting date, duration: 10/2020 - 09/23 (36 months)

Details: Real-world systems are mostly nonlinear in essence. To control these nonlinear systems, linearisation, estimation, or robust techniques could be used to design controllers to suppress the effect of nonlinearities in the system. However, with a nonlinear approach of investigation, the system's dynamics can be elaborated and consequently be used to design more optimum controllers aiming at an exchange between the existing attractors. 

Funding Body: TETFund, Nigeria (PhD studentship) 

Duration: 10/19 - 09/22 (36 months) £101K

Details: Piezo-stack actuated nanopositioning platforms are extremely popular due to their simple construction, mechanical robustness and relatively large motion range. However, their performance is limited by lowly damped mechanical resonances as well as piezo-stack induced hysteresis and creep. We focus on devising new control strategies to overcome these limitations and expand the performance envelope of these nanopositioners to its maximum. 

Funding body, total cost: ESRC, £85,000

Starting date, duration: 09/2021, 48 months

Details: Robots will become increasingly common in many fields of work, from health care to fabrication to logistics. However, because robots look and act differently than other humans, they may not trigger the same mental state attribution processes that otherwise help us to understand what others know and want. This can severely undermine the effectiveness of interactions with robots, especially in fields where anxiety-free cooperation is crucial (e.g., health care). The goal of this interdisciplinary research project is to reveal the extent to which people spontaneously attribute the same mental states to robots as to humans, on which morphological (e.g., eyes) and behavioural features (e.g., biological motion, efficient goal seeking) such attributions rely, and whether manipulating these features can either encourage or discourage people from seeing robots as human-like interaction partners (instead of inanimate objects). It relies on two well-established experimental techniques developed by Professor Patric Bach, which robustly measure two central components of social sense-making in independent research streams: (1) how people predict another actor’s behaviour from the intentions they attribute to them, and (2) how they derive another actor’s knowledge from their particular visual perspective. By varying the robots’ morphological features and behaviours, we will be able to measure these two central types of mental state attribution and link them to explicit ratings of interaction quality and mind perception. In doing so, this project will not only provide new insights into how mental states are attributed to robots – and other humans – but also provide novel methods to give robots characteristics that encourage people to “see” them as endowed with mental states so that they can cooperate with them more confidently. 

Funding body, total cost: UKRI, £65,000

Starting date, duration: 19/2021, 36 months

Details: As robots spread into different areas of human activity, they are breaking through the boundaries of the well-structured industrial environments in which they were first used. This is particularly thanks to the breakthroughs in deep learning, which have given robots an unprecedented level of flexibility and robustness in manipulating objects. However, these advances remain limited to simple actions such as grasping or placing objects under known conditions. Still, we need robots that can perform complex tasks and cope with unforeseen circumstances. Giving robots such capabilities is the goal of this project and of utmost importance for the success of future robotic systems. 

Funding body, total cost: School of Engineering, £85,000

Starting date, duration: 10/2021, 36 months

Details: This research project focuses on the design and development of a compliant, non-invasive and novel wearable device for upper limb rehabilitation. Specifically, this doctoral project is aimed at the development of a variable compliance, lightweight brace that can adapt to the user’s body for comfort, usability and effectiveness. The project will encompass biomechanical modelling to inform the design of the soft robotic device, and fabrication and testing of a prototype device.

Applications for such novel, wearable devices will be in facilitating upper limb rehabilitation and supporting strength augmentation of patients to increase their independence. This device could support rehabilitation of arm function by addressing muscle weakness in the arm, and help prevent undesirable spasticity in muscle related to the pathology of stroke.

Funding body, duration: Commonwealth Scholarship Commission

Duration: 10/22 - 09/23 (12 months) £22.5K

Details: Switched Reluctance Motors (SRM) have emerged as a promising actuator for Electrical Vehicle (EV) applications. However, application-specific design and optimization of SRMs for EV applications is in its infancy. This project aims at exploring optimal SRM designs and establish optimization algorithms for EV-specific SRMs. SRM behaviour under electromechanical braking, thermal profiles, magnetic circuit optimization and efficiency are some of the performance metrics that will be investigated.

Funder: Petroleum Technology Development Fund (PTDF – Nigeria)

Duration: 10/21 - 09/24 (36 months) £110K

Details: Soft-actuators popularly come in two forms – pneumatic soft actuators (PSA) and dielectric elastomer actuators (DEA). While PSA are capable of large deflection and posses high force capability compared to the DEA, the DEA are cleaner to actuate and can deliver highly accurate deflections at micro- and nanometer scales. Nevertheless, modelling and control of both is challenging due to inherent hysteresis and creep nonlinearities. This project aims at establishing accurate control-enabling models and formulating wide-band control strategies that are robust and deliver accurate positioning and trajectory tracking.

Funding body, total cost: Standard Innovation Voucher, £5,000, Knowledge Exchange and Commercialisation (KEC) Award application, £6,000 and under revision by the Medical Device Manufacturing Centre, Edinburgh.

Starting date, duration: 05/2021, 5 months

Details: The Desire Health Mask is a smart mask that shows and monitor the wearer’s likely state of health while also minimising the risk of pathogen exposure via the inhalation route. The state of health of the wearer is tested continuously by the sensory system integrated into the mask. The mask includes pulse rate, oxygen saturation and thermometer to measure the health of the user.  The mask indicates via colour-coded lights (e.g. green/amber/red) the likely state of health of the wearer to people around them, additionally the biometric data of the user is displayed on an OLED screen and externally stored. This project is developed in collaboration with the SME Desire Hygiene and is currently being reviewed by the Medical Device Manufacturing Centre https://mdmc.hw.ac.uk/#infinite

Funding body, total cost: NERC, University of Aberdeen, £3500, £10,000

Starting date, duration: 04/2022, ongoing

Details: Honeybees represent the apex of the evolution of social behaviour, showcasing a wide range of complex social interactions among nestmates that have fascinated scientist (and the general public) for centuries. One of the most intriguing behaviour is the waggle dance communication, a series of stereotyped movements that bee foragers perform in the hive to share information with nestmates about the location (distance and direction from the beehive) of a profitable food source or a new nesting site. This is one of the most complex examples of communication in the animal kingdom, and it has been extensively studied at multiple levels, considering the evolution, the ecological relevance and the biological processes that underpin it – core part of the ongoing research in Dr Fabio Manfredini’s research group at the School of Biological Sciences, University of Aberdeen. The aim of this project is to use robots built to imitate swarm behaviour, called Kilobots, in our custom-built arena to recreate in a controlled environment the movements and the network of social interactions that honeybees perform during the waggle dance.

Funding body, total cost: Pump priming, University of Aberdeen, £10,000

Starting date, duration: 07/2021, ongoing

Details: The aim of this project is to create a multidisciplinary network in the field of human robot interaction (HRI) between  ARMS and Prof. Patric Bach, from the School of Psychology and lead of the Action and Prediction Lab. Robotics and AI are at the forefront of the current forth industrial revolution but in order to bring these technologies to effective fruition it is necessary that that the robot behaves in a way that is easily understandable and predictable for humans,  so that interactions with it foster acceptance and promote effective collaboration. The funds were used to purchase a Pepper robot (in the figure, https://www.softbankrobotics.com/emea/en/pepper). Pepper is a humanoid robot designed for human robot interaction that has successfully been used in the ESRC funded PhD studentship and for dissemination: in the Offer Holder Day in April 2022 and in the Festival of Ideas in May 2022. 

Funding body, total cost: Carnegie Trust Vacation Scholarship, £3500

Starting date, duration: 05/2021, 4 months

Details: Existing state-of-the-art robotic systems can effectively tackle the bin-picking task. However, autonomous robot grasping of unknown objects is often limited to the performance of the Neural Network (NN) governing the system. The success of a never-seen object grasp correlates to the systems’ ability to generalize on new or never-seen data. We developed a strategy for improving Convolutional Neural Network (CNN) performance for object grasping utilizing hyper-parameter tuning and data augmentation. Optimal parameters for the CNN training will yield the most optimal model for the task, thus provide the most optimal and optimized decisions for the bin picking system. We hope that this research adds to the automation of industrial bin picking.

Funding body, total cost: Carnegie Trust Vacation Scholarship, £3500

Starting date, duration: 05/2021, 3 months

Details: Soft robots offer new opportunities because of their compliant physical structure and their wide range of applications.  Currently the development of   such robots is hampered by their low controllability.  One of the main constituents of soft robots are soft actuators. The aim of this project was to improve the control of a non-linear system, the soft actuator, and its interaction with the environment, by training along short-term memory (LSTM) neural network to accurately predict  the actuator’s position in space, its curvature,  and the force applied by its end-effector on an external object. The results are published in the IEEE MMAR 2022 conference proceedings. 

Funding body, total cost: 2020 Carnegie Trust Vacation Scholarship, £3500; 2021 School of Engineering Summer Scholarship, £3000

Starting date, duration: 05/2020, 3 months; 05/2021, 3 months

Details: A two-wheeled self-balancing robot is a statically unstable non-linear system with strong coupling dynamics. Common practices in the development of control systems for such robots are either to linearise the region of application to be used with linear controllers or to use complex nonlinear controllers such as Fuzzy logic, Sliding Mode, and Neural Networks. Nonetheless, in this work, we are proposing a novel to this field concept of switching control that would adjust its approach depending on the evaluation of the current states. This work was presented at the 2021 ICRA conference. A second project used evolutionary algorithms to design control systems with minimal human supervision. In this work, the SMC and switching control are improved using evolutionary algorithms and their performance is assessed in the established as well as in novel experiments. Simultaneously, the effectiveness of evolutionary algorithms coupled with simulation software (CoppeliaSim) for this type of problem is examined. The journal paper describing this work is currently under revision.