Details of Grant 

EPSRC Reference:	GR/K26622/01
Title:	MIME: MAKING IT MOVE EASILY
Principal Investigator:	Patterson, Dr JW
Other Investigators:
Researcher Co-Investigators:
Project Partners:
Department:	School of Computing Science
Organisation:	University of Glasgow
Scheme:	Standard Research (Pre-FEC)
Starts:	01 October 1994	Ends:	30 September 1997	Value (£):	150,060
EPSRC Research Topic Classifications:
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:
Summary on Grant Application Form
This project is about the animation of articulated 3-D human-like figures using a choice-based rather than a skill-based methodology. To the user the choice mechanism looks a lot like that of Todds 'Mutator'. Using these terms the project objectives are: to test the hypothesis that good 3D animations can be evolved in a 'Mutator'-type model to establish a good set of focusing mechanisms to avoid unproductive search sequences to show how to drive 2D animations from previously-derived 3-D animation sequences (and to produce some examples)Progress:Although funding for this project was not available until October 1994 good progress has already been made as work on the topic began at about the time the proposal was submitted. As indicated in the proposal, a concept demonstrator had been built which could show the style of interaction of the user with the articulated figure, although user interaction had little effect on constraining the figures motions. The figure would throw its limbs about the screen in a plausibly realistic manner because the behaviour of the limbs was constrained by C2 continuous dynamic splines, and this underlying machinery filtered out a large set of physically implausible actions from further consideration. Our work has concentrated on an examination of suitable focusing models which could further constrain the possible directions in which the mutation could evolve. We had been using mutations as the main method of obtaining new animations and it is well-known that the usual focusing mechanisms in 'Mutator' are not powerful, and certainly did not seem powerful enough for our application. We have now implemented 'marriages' (between 'parents') using a fully-featured genetic algorithm which has been specially developed for this application. Marriage was always thought to be a preferable model to point mutation because it could exploit libraries of limb-group movements (e.g. legs walking, arms waving) and, although we have not yet implemented libraries, we can now show that genuine convergence is taking place. Unfortunately that convergence is still quite slow if we limit it to marriages between entire sequences (simple animations take 20-40 generations to evolve). However, we now allow incestuous marriages in which parts of sequences can also be marked as preferred (and, if necessary, other parts marked as undesirable). Here convergence is much faster and, depending on choices, can take place in as few as 5 generations. The next stage of our work involves introducing libraries of movements. Genetic algorithms need large populations to work on and the obvious way to provide these is through pre-existing libraries (which we have always intended to have). Our immediate problems will be how to find a start point (use 'Mutator' again) and how to constrain the library search, since this is a finite space we know all about. At present we are working up our workstation and porting pre-existing software to it, including the latest version of our demonstrator.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk Summary Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:
Further Information:
Organisation Website:	http://www.gla.ac.uk