Text

Research Area 2: Control and Robotics

Goal

This research area in multi-robot systems aims to develop sophisticated, efficient, and secure control and robotics systems capable of collaborative operation. This involves creating cutting-edge algorithms for automated planning and distributed control, enabling multiple robotic entities to perform complex tasks in unison with enhanced efficiency and adaptability. Central to this endeavour is ensur ing the cybersecurity of these systems, safeguarding them against potential threats as they become increasingly interconnected and integral to industrial, environmental, and societal applications.

Research Directions

Automated planning of multi-robot systems

We aim to delve into automated planning, employing sophis ticated algorithms and computational models to refine decision-making in multi-robot systems. We aim to leverage cutting-edge techniques to boost efficiency, minimise redundancy, and enhance performance across various operational environments.

Distributed control and coordination of multi-robot systems

This research area focuses on developing in novative distributed control and coordination mechanisms. We aim to establish a framework where robots autonomously adapt to evolving situations, share information effectively, and collaboratively accomplish intricate tasks. This approach seeks to improve robotic cooperation and intelligence

Cybersecurity in control systems

As integrating robotics and automation systems with digital technologies intensifies, so does the importance of safeguarding these systems from cyber threats. This direction will focus on cybersecurity within control systems. We are committed to strengthening the resilience of multi-robot systems against potential vulnerabilities, ensuring data integrity, confidentiality, and system availability. By implementing cybersecurity measures, we aim to build trust in the reliability and security of automated solutions.

Background

In the rapidly evolving landscape of technology and automation, the advent and devel opment of multi-robot systems stand as a beacon of innovation and efficiency. As industries become more reliant on automation, managing multiple robotic entities cohesively and efficiently becomes im perative. This is not merely a pursuit of operational efficiency but a fundamental shift in how complex tasks and processes are approached and executed. From manufacturing floors to intricate logistics operations, multi-robot systems promise to enhance throughput and precision and bring scalability and adaptability to operations.

Yet, as these systems become more interconnected and data-dependent, they also become vulnerable to new threats. In this context, cybersecurity is not just a technical necessity but a cornerstone for trust and reliability. Protecting these systems from cyber threats is paramount to safeguarding sensitive data and maintaining the operational integrity of these complex networks. The pursuit of robust cyber security measures within multi-robot systems is thus a critical endeavour that underpins the successful integration of these technologies into the fabric of society.

Moreover, the shift towards distributed control in multi-robot systems marks a departure from traditional centralised models. This paradigm shift is important to overcome the inherent limitations of centralised systems, such as scalability bottlenecks and inflexibility. Distributed control makes resilience and adapt ability essential in rapidly changing environments, ensuring that multi-robot systems are efficient and robust in the face of unforeseen challenges.

Methods

The research projects in RA2 employ a comprehensive methodology to address key questions using rigorously defined techniques.

Our methodology is anchored in sophisticated mathematical models for designing automated planning algorithms, distributed control systems, and cybersecurity in automation. For multi-robot automated planning, these models enable the development of efficient, optimal algorithms. We aim for a balance between precision and speed, integrating artificial intelligence, machine learning, and optimisation the ory. We're exploring evolutionary algorithms like Genetic Algorithms, Particle Swarm Optimisation, and Simulated Annealing, and investigating Large Language Models (LLMs) for automated planning through user input.

In distributed control, we create agent-based models to understand robot interactions and enhance coordination. We use tools like MATLAB, ROS, JADE, and Gama to test our control mechanisms in simulated multi-agent environments, assessing scalability and coordination efficiency.

For cybersecurity, we develop real-time model-based anomaly detection algorithms and enhance them with machine learning, improving the security of control and robotic systems.

Expected results

In our project on automated planning for multi-robot systems, we aim to develop novel and sophisticated algorithms, focusing on optimising tasks, resources, and motion. This includes robust mathematical models to understand complex robot interactions and comprehensive simulations to evaluate algorithm performance in various domains.

We will apply our research in real-world scenarios, partnering with industries to showcase its practi cality. We focus on decentralised control in multi-robot systems, shifting from traditional centralised methods to more efficient, large-scale coordination. We will develop agent-based models to simulate individual robot behaviors, aiding in designing innovative coordination strategies. These models will be tested in simulated environments to assess scalability and coordination effectiveness.

Enhancing cybersecurity in control systems is a key aspect. We will develop real-time anomaly detec tion algorithms and control strategies to secure multi-robot systems against cyber threats, integrating machine learning for swift risk identification. Our cybersecurity frameworks, tailored for multi-robot environments, will ensure system integrity and reliability. We plan empirical studies to ensure our cy bersecurity measures boost system performance without compromising efficiency or decision-making.