An automatic system to translate a computer model of machinery into a diagram showing how it works has been developed by a UCL-led team of computer scientists.
‘Illustrating How Mechanical Assemblies Work’, featured in the January 2013 Communications of the ACM Research Highlights section, describes a method which analyses the potential motions a mechanical model can undergo, adds annotations and then produces animations and sequential images explaining the functioning of the assembly. The researchers hope it will help students and users of technical documentation understand the operation of many kinds of mechanical assemblies, and demonstrate its functioning on various models, including mechanisms originally sketched by Leonardo da Vinci.
The research was carried out by Dr Niloy Mitra (UCL Computer Science), his postdoctoral students and colleagues from Adobe (Dr. Wilmot Li) and University Of California Berkeley (Prof. Maneesh Agrawala), and an earlier version was presented at ACM SIGGRAPH 2010.
The method relies on identifying parts in the assembly based on their geometry and symmetries. Their motion and potential to interact with each other is identified automatically – the system is capable of recognising cam and crank mechanisms, axles, belts, and several kinds of gears. Meanwhile, the user picks out the part that receives external input – the driver – and the direction in which it moves, which can then be propagated through the assembly.
The operation of the machinery can then be conveyed to the user through animations and sequential images highlighting key moments, such as when parts change direction. The explanation is made clearer through the addition of directional arrows and the highlighting of active parts.
Lead author on the paper, Dr Niloy Mitra, explained the motivation behind his work:
“As tablet devices become more common, we think there is no reason for books to be static. Using our techniques, many textbook images can become interactive. Young people and students of any age can ‘play’ with the figures and learn how parts interact and mechanisms function.”
Future work will focus on extending this method to more complicated mechanisms including those that incorporate fluids, and also visualizing the action of forces in the mechanism.
See a video about the technique here: