Laban Movement Analysis (LMA) is a popular method for analyzing, interpreting, and communicating movement qualities introduced by Rudolf Laban. Originating from the exploration of dance movement, it has been used extensively to characterize movement not just in dance but in sports, theatre, and film. More recently, LMA has inspired AI methods to automatically characterise human movement and generate animations for virtual agents and humanoid robots.
The generation of expressive behaviour for virtual agents and robots has been dominated by inscrutable deep learning end-to-end models, with some approaches enabling some degree of control over this generation with high-level variables. This research project aims to explore LMA as a general and intuitive representation for perceiving, generating and controlling artificial agents’ movements and expressive qualities, and assessing its applicability not just in the context of dance, but for any context featuring expressive movement. The project will focus on interactive settings, such as interactive performance with artificial agents, the generation of LMA-based feedback to performers, LMA instruction, etc.
Proposed Methods
We will adapt LMA as a tool for modeling movement computationally, which may include the definition and training of machine learning models from LMA-annotated datasets for learning a representation of movement primitives corresponding to the dimensions of LMA, i.e., body, shape, effort, space, and their sub-dimensions. Modern methods, such as Graph Convolutional Networks – well adapted to representing the skeleton structure of human animations – may be used. These learned representations and models will be deployed in interactive tasks for experimental studies with the goal of assessing LMA as an intuitive scheme for controlling artificial agents’ movements or providing feedback on human movement qualities. Comparisons will be realised between populations of subjects trained in LMA (such as Laban-trained professional dancers) with naive participants to assess its generalisability and intuitiveness.
Likely Outputs
This project will lead to the development of a general and intuitive representation for controlling artificial agents’ movements and expressive qualities based on Laban Movement Analysis (LMA). The project will also generate insights into the controllability of computational methods leveraging artificial intelligence to interact with expressive artificial agents.
Possible Impact
This project has the potential to provide a radically new and intuitive way to control and interact with artificial agents. This could have applications in varied contexts, such as human-agent or human-robot interaction in social or performance settings, for dance and movement training, or for movement assessment, e.g., rehabilitation.
Alignment with Industrial Interests
The project will bridge the gap between the performing arts and the technology industry, providing a glimpse into the future of the performing arts.
Brief Timetable
First, modern methods for characterising and generating expressive movement will be reviewed, with a focus on the use of LMA as a computational representation. This will include the identification of relevant expressive movement datasets. The review will also consider how LMA was adapted in computational methods to discriminate or generate movements.
The project will then move on to designing a LMA-inspired computational representation model, possibly following GCNs or alternative neural network architectures, which will be used as the basis for expressive animation generation. The project will then proceed to deploy this model in a first interactive setting, such as virtual choreographic control of artificial agents, evaluating its use by naive or trained LMA subjects in terms of intuitiveness, expressiveness, and generative control. Other settings will then be explored, such as performance or conversational settings.
Eligibility
Applicants must have or expect to obtain the equivalent of a 1st or 2:1 degree in any subject relevant to the CDT including, but not limited to, computing science, psychology, linguistics, mathematics, sociology, engineering, physics, etc.
Applicants will be asked to provide two references as part of their application.
Funding
Funding is available to cover the annual tuition fees for UK home applicants, as well as an annual stipend at the standard UKRI rate (currently £17,668 for 2022/23). To be classed as a home applicant, candidates must meet the following criteria:
- Be a UK National (meeting residency requirements), or
- Have settled status, or
- Have pre-settled status (meeting residency requirements), or
- Have indefinite leave to remain or enter.
As per UKRI funding guidelines, up to 30% of studentships may be awarded to international applicants who do not meet the UK home status requirements. Funding for successful international students will match that of home students and no international top-up fees will be payable.
Continue with Facebook
