Player Activity Tracking App: Improving Athlete Performance with Machine Learning Powered Predictions
JavaScript is disabled in your browser.
Please enable JavaScript settings in your browser to allow this website work correctly.
Customer
One of the world’s top semiconductor chip makers.
Problem
Started as a POC project to assess the potential of a homegrown technology, the initiative has spun into a complete solution implementation. Impressed with the level of skills across the target domains spanning low-level programming, mobility solution along with hardware to software communication, the customer tapped Softeq’s expertise again tasking the team with transforming the prototype mobile app into a fully functional demo of its flagship product capabilities.
The demo app was to showcase the wide range of features of the customer’s innovative platform powering a puck-like device created for action sports.
Solution
The device with an in-built advanced compute module, 9-axis sensor, and GPS system can be mounted onto a gamut of sports equipment, including snowboard, surfboard, skateboard, BMX bike, wakeboard, basketball, and more. Paired with a specially developed app, the complete solution is set to enable professional and casual sports enthusiasts to improve their technical skills by tracking the trick-making progress.
Using radio or BLE frequency technology, the mobile app connects wirelessly to the sensor that works with accelerometers and gyroscopes providing for the following capabilities:
- Recording a full HD video of the trick, while retrieving data from the sensor unit
- Showing video preview with:
- - Charts generated based on the data retrieved from the sensor unit
- - Real-time speed parameters based on a customized speed calculation algorithm
- - Real-time 3D spinning model
- - Trick and move recognition
- Exporting video into app, while adding an overlay of charts, patterns, and speed values
- Supporting external cameras
Challenge
When customizing the speed calculation algorithm, the team worked around the problem of processing big chunks of data retrieved from multiple sensor units using a complex filtering mechanism. The team also had to come up with solutions for overcoming the UWP technical limitations.
Methodology
The team employed an iterative method developing the solution incrementally.
Results
While still in the prototype state, the solution was test-driven in real world environment by a pro surfer, multiple time big wave competition honoree, who confirmed the viability of the system under development. The extended version of the demo app was showcased during a public presentation at a development conference organized by the customer.
