At-Home Gym is easier and more accurate with AI & Signal Processing

Read your movements and achieve your fitness goals with AI, the perfect at-home gym. Learn more about our solution.

Data Science

Embedded Computing

Signal Processing

Industry

Beauty & Wellness

Service

Scale your team

About Tonal

Tonal is a smart home gym that combines strength training equipment with technology. It uses magnetic weights and cable machines to offer a variety of exercises and personalized workout programs.

Founded in San Francisco in 2015 and raised more than 450M USD, its concept revolutionizes fitness and strength training with patented digital weight, machine learning, and expert coaching.

“Tonal Trainer,” an at-home gym solution, is one of the most space-efficient in the market. It can replace traditional gym equipment like dumbbells and barbells and provide feedback on your form during workouts.

The Challenge

The Tonal workout system uses a large touchscreen and controllable arms to provide resistance training at home. Sensors track the user's movements, but the challenge lies in combining data from these sensors that do not necessarily record information at the same rate. This requires unique algorithms that can handle this irregular data and still provide real-time feedback during the user's workout, all while being efficient on the device.

The Solution

Our team of engineers has developed a custom sensor fusion algorithm. This algorithm fuses the data sources available in the trainer device to derive estimates of hand trajectories in space. The developed solution has been ported for real-time execution in the available hardware, which has constrained resources (memory and computation time).

A thorough analysis of the sensor streams was performed. This includes estimating actual sampling rates from different sensors and defining an appropriate preprocessing pipeline (e.g. resampling, synchronization, bias removal) for data sanitization before its use in the sensor fusion algorithm.