Within this field, the ARMS group research themes are in the development of true Artificial Intelligence methods, in particular in the areas of automated AI (where the AI can learn the important features of a data problem in an automated manner, without human involvement and necessity for parameter tuning), autonomous learning (where the AI method can learn and adapt to any given environment, and changes within it), and explainable AI (where the AI method can explain fully how it reaches its decisions).  These areas have all been recognised as the next main challenges for AI.

These techniques have been successfully applied to a number of different fields, including robotics, econometrics (the study of financial markets), bioinformatics (in particular genomic and proteomic analysis), engineering problems and the analysis of seismic data and the integration of AI and virtual reality.

Recently a member of the ARMS co-developed a system (utilising image processing and AI) capable of generating music in real-time whilst the artist painted to produce a novel live music performance!  An example of the artwork used by the AI to produce music can be seen on the left.

Flexible robotics research within the ARMS group focuses on creating robots with soft and flexible materials instead of rigid ones. These robots are designed to be more adaptable and resilient, able to navigate complex environments and interact with objects and people without causing harm. Flexible robots have applications in cases where there is a need for less energy consumption or an increase in operation speed compared to traditional rigid robots.

Origami robots combine the principles of origami, the ancient art of paper folding, with robotics. The goal of origami robotics is to create robots that can be easily transported and deployed in various environments, such as disaster areas or space missions. These robots are designed using origami techniques to make them lightweight, compact, and easy to deploy.

The combination of both origami robotics and flexible robots is pushing the boundaries of what traditional robots can do, with applications in fields such as healthcare, where they can be used for minimally invasive surgeries.

Soft Robotics research within the ARMS group focuses on the use of variable compliance within embodied intelligence to design innovative mechanisms and robotic platforms with inherent safety in human-robot interaction and adaptive robot grasping.  Bioinspired variable compliance is also exploited to engineer novel and effective systems for real-world environments, especially for healthcare and assisted living.  

This research is built on the design of variable compliance actuators that can readily adjust to environmental changes and unstructured tasks, offering an effective paradigm for adaption to ever-changing real-life, unstructured environments.  The challenging task of controlling soft and variable compliance actuators is tackled utilising innovative solutions based on neural networks and bioinspired algorithms.  The union of innovative actuator design and bioinspired control schemes can be applicable to numerous challenges.  These can range from rehabilitation and disability assistance to effective grasping.

Human-robot interaction (HRI) is the research of interactions between humans and robots.  This is a highly multidisciplinary field with links to human-computer interaction, artificial intelligence, robotics, natural language understanding, and psychology.  As robots are becoming more advanced, and are emerging from the laboratories of the world into the real world the way in which humans and robots can work together successfully will be incredibly important.  The robots must behave in a way that humans are comfortable with while being able to complete the task at hand.  Ultimately a better understanding of human perception is necessary for designing robots to fit this critera.

Currently, the ARMS group's HRI research is focused on designing novel robot control algorithms to promote various mental state attributions in humans (for example efficient goal-seeking and sensing/avoidance of obstacles).  The ARMS research group is working on a joint ESRC funded project with Professor Patric Bach's Action Prediction Laboratory.  At present this project is focused on developing a more thorough understanding of the multimodal processing of the human perceptual and cognitive system.  The first part of this is developed motion planning algorithms for the limbs of our robotic platform that are able to emulate organic human motion for a quick grasping task.  It is theorised that enabling robots to move in an organic way will affect the perception of them by humans positively.  Ultimately it is hoped that this research can evolve into achieving fluid cooperation in complex tasks between humans and robots.  

There are three key research themes:

1. Kinematic and Dynamic prototyping and analysis: This involves detailed investigation of the system behaviour in both time and (if applicable) frequency domain. Key outcomes are the complete understanding of the current performance envelope and possible performance improvements. A wide array of sensors, amplifiers, Digital Signal Analyzer, High-bandwidth Digital Oscilloscopes, LabView installations as well as the state-of-the-art dSPACE MicroLab Box are available for such investigations. 

2. System Modeling: This involves establishing and validating a mathematical model that captures the system dynamics within the bandwidth-of-interest. Integer- / fractional-order linear / nonlinear differential equations as well as data-based models are employed. Access to the high-density computing facility is also provided for high-dimensional data processing and analysis.

3. Control development: The aim is to find the simplest possible solution that extends the performance boundaries of a given system, in a stable and repeatable fashion. This ensures hardware simplicity, low computational load, easy tunability and transparent signal flow, and consequently, allows for the control strategies to be implemented in a real-world setting, not just on a lab-based test bench. Robustness to parameter uncertainty, long-term stability and graceful performance degradation due to ageing, external unplanned disturbances and deviations from standard operation are embedded. After extensive simulations using the MATLAB / SIMULINK environment, the designed control algorithms are experimentally validated using NI LabView or dSPACE.

Systems we have worked with: Piezoelectric tube and platform-type nanopositioners, Hexapods, drill-strings, flexible robotic manipulators, soft robotic actuators, self-balancing robots, high-performance piezo-actuators, switched reluctance motors for EV, non-destructive testing kits, structural health monitoring systems, active vibration control / isolation systems and MEMS.

Email s.aphale@abdn.ac.uk to discuss research collaborations, joint research projects, PhD projects, invited lectures and short courses within this area.