The HPC Behind the Product Series
April 15, 2021
Shelter-in-place orders and gym closures forced most of us to seek alternative workout options to stay fit during the COVID-19 pandemic. It's no surprise that the home gym revolution got a popularity boost and fit our new reality perfectly. We discussed the Peloton in a recent blog and are excited to highlight the HPC behind another one of the slickest solutions for an inconspicuous home gym, MIRROR, in this blog post.
The hardware is a large glass mirror about 4½ ft tall and almost 2 ft wide. Being only 1½ inches thick, MIRROR can be mounted or leaned against a wall using anchors. Behind the glass is an LCD screen powered by a series of integrated circuits that make up the core of the product’s artificial intelligence engine. Advanced camera technology and strategically placed microphones allow MIRROR to interact with the user and deliver personalized workout content. Don’t worry, the camera and the microphone need your permission to be turned on.
The Science and Technology
The science and technology employed to achieve such a feat are impressive. Computer chips need proper ventilation and cooling to function at high efficiency. Computer modeling and simulation of the layered structure of MIRROR were employed to optimize the thermal performance of MIRROR.
We will go over several things that the MIRROR designers may have optimized using computational methods, including junction temperatures of the electronics, touch temperature of the glass cover, and heating of the camera.
Computational fluid dynamics (CFD) analysis of a layered LCD with a camera, as shown below, gives you an idea of how the designers and developers used cloud computing to design and develop MIRROR to achieve optimal performance.
Replication of CFD analysis of buoyancy-driven airflow around MIRROR at steady state
How does MIRROR gather and track biometric data?
To gather your biometric data, MIRROR connects to your heart rate monitor via Bluetooth to deliver real-time workout personalization. Suppose you do not have a heart rate monitor or a smartwatch during your workout sessions. In that case, MIRROR estimates your calorie count based on an algorithm that incorporates your stats like your weight, age, max heart rate, and the intensity of the exercises in your workout.
What algorithms and AI processes were likely involved in delivering in-workout adjustments based on your goals, preferences, and personal profile?
MIRROR continuously improves and helps you achieve your workout goals by using a neural network and artificial intelligence engine. The AI software used by MIRROR is probably very similar to AI software such as Adaptive Coach that the startup Adaptive uses for their popular fitness app. The coach’s neural network is a series of algorithms that attempt to recognize underlying relationships in a set of data through a process that mimics how the human brain operates. Adaptive Coach software has been trained (learning from existing data with already verifiable outcomes) using data collected from millions of workouts that the company streams worldwide. The image below shows the basic diagram of a neural network used to deliver personalized workouts to fitness consumers.
An example of a neural network diagram used for a personalized workout with node weight and bias hidden
From a scientific perspective, MIRROR’s neural network relies on several mathematical concepts such as vectors, layers, and regression.
- A neural network uses your personalized workout information and biometrics to predict what your workout routine should be in order to reach a specific goal.
- Once the comparison is complete, the network adjusts its internal state to improve the next prediction and offers you an optimal workout regime.
AI and simulation have become powerful tools used in developing innovative products in record time. Home gym equipment like MIRROR and Peloton offers a solution towards overcoming the barriers to physical activity that many of us encounter. And while a typical workout routine only represents about 2% of someone’s week, many products compete for that share.
Stay tuned for the next article in our HPC Behind the Product series, highlighting the HPC involved in designing a YETI tumbler.
Disclaimer: We are not affiliated, associated with, or in any way represent MIRROR.