Evolutionary algorithms (EAs) have been applied to solve many stationary problems. However, real-world problems are usually more complex and dynamic, where the objective function, decision variables, and environmental parameters may change over time. In this project, we will investigate novel EA approaches to address dynamic optimisation problems (DOPs), a challenging but very important research area. The proposed research has three main aspects: (1) designing and evaluating new EAs for DOPs in collaboration with researchers from Honda Research Institute Europe, (2) theoretically analysing EAs for DOPs, and (3) adapting developed EA approaches to solve dynamic telecommunication optimisation problems. In this project, we will first construct standardised, both discrete and continuous, dynamic test environments based on the concept of problem difficulty, scalability, cyclicity and noise of environments, and standardised performance measures for evaluating EAs for DOPs. Based on the standardised dynamic test and evaluation environment, we will then design and evaluate novel EAs and their hybridisation, e.g., Estimation of Distribution Algorithms (EDAs), Genetic Algorithms, Swarm Intelligence and Adaptive Evolutionary Algorithms, for DOPs based on our previous research. A guiding idea here is to improve EA's adaptability to different degrees of environmental change in the genotypic space, be it binary or not. Systematically and adaptively combining dualism-like schemes for significant changes, random immigration-like schemes for medium changes, and general mutation or variation schemes for small changes, is expected to greatly improve EA's performance in different dynamic environments. And memory schemes can be used when the environment involves cyclic changes. In order to better understand the fundamental issues, theoretical analysis of EAs for DOPs will be pursued in this project. We will apply drift analysis and martingale theory as the starting point to analyse the computational time complexity of EAs for DOPs and the dynamic behaviour of EAs for DOPs regarding such properties as tracking error, tracking velocity, and reliability of arriving at optima. Based on the above EA design, experimental evaluation, and formal analysis, we will then develop a generic framework of EAs for DOPs by extracting key techniques/properties of efficient EAs for DOPs and studying the relationship between them and the characteristics of DOPs being solved with respect to the environmental dynamics in the genotypic space. Another key aspect of this project is to apply and adapt developed EAs for general DOPs to solve core dynamic telecommunications problems, e.g., dynamic frequency assignment problems and dynamic call routing problems, in the real world. We will closely collaborate with researchers from British Telecommunications (BT) to extract domain-specific knowledge and model dynamic telecommunication problems using proper mathematical and graph representations. The obtained domain knowledge will be integrated into our EAs for increased efficiency and effectiveness. All algorithms and software developed in this project will be made available publicly to benefit as many users as possible, whether they are from academe or industry.

The overall theme of our proposed doctoral programme is ECOLE: Experience-based COmputation: Learning to optimisE. It seeks novel synergies between nature inspired optimisation and machine learning to address new challenges that arise in industry due to the increasing complexity of products, product development and production processes. The unique aspect of ECOLE is to study and capture the notion of experience that is associated with expert engineers, who have worked on complex optimisation tasks for a certain time, in a computational framework composed of machine learning and optimisation strategies. We aim at developing cutting-edge optimisation algorithms that can continuously accumulate experience by learning from development projects both over time and across different problem categories. The more such algorithms are used for different optimisation problems, the better they become since their accumulated experience increases. The Consortium consists of two world-leading universities, the University of Birmingham (UK) and the University of Leiden (The Netherlands), both in the top 150 in the 2016-17 Times Higher Education World University Rankings, and two innovative companies, Honda Research Institute Europe GmbH (Germany) in the automotive sector and NEC Europe Ltd (UK) in the ICT sector. All have world-leading research groups with complementary expertise that support ECOLE. ECOLE fills an urgent need in Europe for highly skilled optimisation and machine learning experts who have first-hand industrial experiences allowing sustainable know-how growth for solving future challenges. Its entire training programme is centred around a set of novel research projects proposed for early stage researchers (ESRs), complemented by domain knowledge training, hands-on engineering training and transferable skill training. ESRs will spend 50% of their time in the non-academic beneficiaries and be trained in different academic environments and industrial sectors.

One of the major drivers of research in the area of humanoid robotics is the desire to achieve motions involving close contact between robots and the environment or people, such as while carrying an injured person, handling flexible objects such as the straps of a knapsack or clothes. Currently, these applications seem beyond the ability of existing motion synthesis techniques due to the underlying computational complexity in an open-ended environment. Traditional methods for motion synthesis suffer from two major bottlenecks. Firstly, a significant amount of computation is required for collision detection and obstacle avoidance in the presence of numerous close contacts between manipulator segments and objects. Secondly, any particular computed solution can easily become invalid as the environment changes. For instance, if the robot were handling an object such as a knapsack, even small deformations of this flexible object and minor changes in object dimensions (e.g., between an empty bag and a stuffed bag) might require complete re-planning in the current way of solving the problem. Similar issues arise in the area of computer animation, where there is a need for real-time control of characters - moving away from static sequences of pre-programmed motion. Although it may seem that this world is much more contained, as it is created by an animation designer, there is in fact a strong desire to create games and simulation systems where the users get to interact with the world continually and expect the animation system to react accordingly. This calls for the same sort of advances in motion synthesis techniques as outlined above.The fundamental problem lies in the representation of the state of the world and the robot. Typically, motion is synthesizes in a complete configuration or state space represented at the level of generalized coordinates enumerating all joint angles and their 3D location/orientation with respect to some world reference frame. This implies the need for large amounts of collision checking calculations and randomized exploration in a very large search space. Moreover, it is very hard to encode higher level, semantic, specifications at this level of description as the individual values of the generalized coordinates do not tell us anything unless further calculations are carried out to ensure satisfaction of relevant constraints. This is particularly inconvenient when searching for a motion in a large database. The focus of this research is to alleviate these problems by developing methods that exploit the underlying topological structure in these problems, e.g., in the space of postures. This allows us to define a new search space where the coordinates are based on topological relationships, such as between link segments. We refer to this space in terms of 'topology coordinates'. In preliminary work, we have shown the utility of this viewpoint for efficient motion synthesis with characters that are in close contacts. We have also demonstrated that this approach is more efficient for categorizing semantically similar motions. In this project, we will develop a more general framework of such techniques that will be applicable to a large class of tasks carried out by autonomous humanoid robots and virtual animated characters. Moreover, we will implement our techniques on industrially relevant platforms, through our collaborators at Honda Research Institute Europe GmbH and Namco Bandai, Japan.

Evolutionary algorithms (EAs) have been applied to solve many stationary problems. However, real-world problems are usually more complex and dynamic, where the objective function, decision variables, and environmental parameters may change over time. In this project, we will investigate novel EA approaches to address dynamic optimisation problems (DOPs), a challenging but very important research area. The proposed research has three main aspects: (1) designing and evaluating new EAs for DOPs in collaboration with researchers from Honda Research Institute Europe, (2) theoretically analysing EAs for DOPs, and (3) adapting developed EA approaches to solve dynamic telecommunication optimisation problems. In this project, we will first construct standardised, both discrete and continuous, dynamic test environments based on the concept of problem difficulty, scalability, cyclicity and noise of environments, and standardised performance measures for evaluating EAs for DOPs. Based on the standardised dynamic test and evaluation environment, we will then design and evaluate novel EAs and their hybridisation, e.g., Estimation of Distribution Algorithms (EDAs), Genetic Algorithms, Swarm Intelligence and Adaptive Evolutionary Algorithms, for DOPs based on our previous research. A guiding idea here is to improve EA's adaptability to different degrees of environmental change in the genotypic space, be it binary or not. Systematically and adaptively combining dualism-like schemes for significant changes, random immigration-like schemes for medium changes, and general mutation or variation schemes for small changes, is expected to greatly improve EA's performance in different dynamic environments. And memory schemes can be used when the environment involves cyclic changes. In order to better understand the fundamental issues, theoretical analysis of EAs for DOPs will be pursued in this project. We will apply drift analysis and martingale theory as the starting point to analyse the computational time complexity of EAs for DOPs and the dynamic behaviour of EAs for DOPs regarding such properties as tracking error, tracking velocity, and reliability of arriving at optima. Based on the above EA design, experimental evaluation, and formal analysis, we will then develop a generic framework of EAs for DOPs by extracting key techniques/properties of efficient EAs for DOPs and studying the relationship between them and the characteristics of DOPs being solved with respect to the environmental dynamics in the genotypic space. Another key aspect of this project is to apply and adapt developed EAs for general DOPs to solve core dynamic telecommunications problems, e.g., dynamic frequency assignment problems and dynamic call routing problems, in the real world. We will closely collaborate with researchers from British Telecommunications (BT) to extract domain-specific knowledge and model dynamic telecommunication problems using proper mathematical and graph representations. The obtained domain knowledge will be integrated into our EAs for increased efficiency and effectiveness. All algorithms and software developed in this project will be made available publicly to benefit as many users as possible, whether they are from academe or industry.

Evolutionary algorithms (EAs) have been applied to solve many stationary problems. However, real-world problems are usually more complex and dynamic, where the objective function, decision variables, and environmental parameters may change over time. In this project, we will investigate novel EA approaches to address dynamic optimisation problems (DOPs), a challenging but very important research area. The proposed research has three main aspects: (1) designing and evaluating new EAs for DOPs in collaboration with researchers from Honda Research Institute Europe, (2) theoretically analysing EAs for DOPs, and (3) adapting developed EA approaches to solve dynamic telecommunication optimisation problems. In this project, we will first construct standardised, both discrete and continuous, dynamic test environments based on the concept of problem difficulty, scalability, cyclicity and noise of environments, and standardised performance measures for evaluating EAs for DOPs. Based on the standardised dynamic test and evaluation environment, we will then design and evaluate novel EAs and their hybridisation, e.g., Estimation of Distribution Algorithms (EDAs), Genetic Algorithms, Swarm Intelligence and Adaptive Evolutionary Algorithms, for DOPs based on our previous research. A guiding idea here is to improve EA's adaptability to different degrees of environmental change in the genotypic space, be it binary or not. Systematically and adaptively combining dualism-like schemes for significant changes, random immigration-like schemes for medium changes, and general mutation or variation schemes for small changes, is expected to greatly improve EA's performance in different dynamic environments. And memory schemes can be used when the environment involves cyclic changes. In order to better understand the fundamental issues, theoretical analysis of EAs for DOPs will be pursued in this project. We will apply drift analysis and martingale theory as the starting point to analyse the computational time complexity of EAs for DOPs and the dynamic behaviour of EAs for DOPs regarding such properties as tracking error, tracking velocity, and reliability of arriving at optima. Based on the above EA design, experimental evaluation, and formal analysis, we will then develop a generic framework of EAs for DOPs by extracting key techniques/properties of efficient EAs for DOPs and studying the relationship between them and the characteristics of DOPs being solved with respect to the environmental dynamics in the genotypic space. Another key aspect of this project is to apply and adapt developed EAs for general DOPs to solve core dynamic telecommunications problems, e.g., dynamic frequency assignment problems and dynamic call routing problems, in the real world. We will closely collaborate with researchers from British Telecommunications (BT) to extract domain-specific knowledge and model dynamic telecommunication problems using proper mathematical and graph representations. The obtained domain knowledge will be integrated into our EAs for increased efficiency and effectiveness. All algorithms and software developed in this project will be made available publicly to benefit as many users as possible, whether they are from academe or industry.