It’s no secret to anyone who watches competitive diving, especially on the Olympic level, that the sport involves a great deal of precision and grace. It’s also difficult for the diver to replicate that precision several times over during an event.
Dive Mechanic, a new tool created by Australia’s Commonwealth Scientific and Industrial Research Organization (CSIRO) for Diving Australia, is designed to help coaches and divers visualize the aerial portion of the dive and make necessary adjustments in technique.” The software — recently recognized at the AIIA iAwards — models in 3D the dynamics of the dive, which is digitized from video footage and a laser body scan of the diver.
According to Raymond Cohen, a research scientist with CSIRO’s Data61 division, these visualizations allow for the testing of a number of important conditions, including take-off velocity and rotational speed. The software also enables coaches and trainers to experiment with different timings, and how earlier or later movement could affect the technique. With all the experimentation in variables that is possible, Cohen said the software could be most useful for helping coaches and divers understand the consequences of different changes in technique.
Perhaps Dive Mechanic’s first practical use was in the lead-up to the 2016 Olympics in Rio de Janeiro. Diving Australia and the Australian Institute of Sport were given visualizations of diving pairs for the synchronized diving competitions, and Dive Mechanic modeled the dynamics of the different pairs. Cohen said that the Australian coaches used the software, and it showed: Australia took home the bronze in the synchronized women’s three-meter springboard event.
“We developed a platform. How the coaches and athletes use this sort of tool, and how they integrate it into their training, is up to them,” Cohen said.
Get The Latest Sports Tech News In Your Inbox!
The models aren’t extremely difficult to compute, and Cohen’s team tries to work as generally as possible; the model for diving could also be used for swimming and even horse racing. Data61 has had a 10-year relationship with Swimming Australia, helping to optimize elite swimming technique, and has completed several other projects with Diving Australia.
Yet Dive Mechanic doesn’t have immediate “poolside” availability. It takes 30 minutes to an hour for a person to examine video of a single dive and render it digitally, meaning the technology is still only semi-automated. Cohen said that one day when the technology is fully automated, real-time modeling could be more realistic.
Currently, Dive Mechanic has a database of approximately 60 dives. With thousands of dives, the modeling and analysis could help analysts and coaches understand the technique of a really good dive. Cohen looks forward to a system that can produce clean data that works seamlessly with the models, which he says would allow the creation of even more complicated models. He also aspires to connect Dive Mechanic to other modeling software that can analyze the splash from a dive.
“It’s good to be able to develop some algorithms, and then be able to deliver those to a client in a useful form,” Cohen said. “This project is a good example of that.”