How cameras, computer vision, and patented technologies are turning treadmills into running coaches.
After a run on the treadmill, we're used to seeing a familiar set of numbers; distance, time, average speed, and calories burned. But there's one question, those numbers can't quite answer.
How well was I actually running?
Two people can run at exactly the same speed while moving in very different ways. One might take quick, short steps, while another runs with longer strides. Some runners bounce more than others, and running from can gradually change as fatigue sets in. Now a new generation of treadmills is beginning to look beyond speed and distance - using cameras to analyze how we move.
In September 2026, Peloton introduced Tread Vision, a camera-based feature that tracks running metrics such as cadence, stride length, and vertical oscillation in a real time. After a workout, it provides a Running Efficiency Score along with personalized tips to help improve running form.
Unlike wearable devices that rely on sensors attached to the body, this technology observe the runner from the outside. So how does a camera turn a video of someone running into meaningful data - and eventually, coaching advice?
Let's explore the patented technologies that help explain how it works.
The patents featured in this article are presented as examples of technologies related to camera-based running analysis and coaching. They are not identified as patents used in Peloton's Tread Vision.
1. First, the Camera Learns to Read Your Movements.
To a camera, a runner is initially just a sequence of images. Before a computer can analyze running form, it needs to identify the person in those images and determine how different parts of the body are moving. This is where technologies such as pose estimation and body tracking come into play.
A patent from Blue Goji describes a system that combines RGB cameras, infrared depth sensing technology, and other sensors to track body movements. By identifying the relative positions and orientations of body parts, the system can run visual information to movement data. In other words, it transforms a moving person on screen into something a computer can measure and analyze.
 |
wipsglobal.com
US 20210069547 A1
Apparatus for natural torso and limbs tracking and feedback for electronic interaction |
Interestingly, a camera doesn't necessarily need a clear view of the runner's feet to detect running rhythm. An earlier patent proposed estimating cadence by analyzing the subtle, repetitive up-and-down movements of a runner's head and torso. The key isn't simply capturing an image. It's recognizing how the body changes position over time.
2. Turning Movement into Running Metrics
Once the system can track body movement, the next step is turning that information into numbers.
Running is repeating cycle a foot lands, supports the body, pushes off, and moves through the air before the next landing. By analyzing the time and position of these movements, a system can calculate a range of running metrics.
Some of the most useful include:
· Cadence: The number of steps taken per minute
· Stride length: The distance covered during a complete stride
· Vertical oscillation: How much the body moves up and down while running
· Contact time: How long a foot remains in contact with the running surface
· Flight time: How long both feet are off the ground
A patent from AiKynetix offeres a particularly good example of how this works.
 |
wipsglobal.com
US 2024-0428621 A1
Systems and methods for measurement and analysis of human biomechanics with single camera viewpoint |
The technology uses a single camera to capture a runner's movement. Computer vision identifies key body positions in the video, while a biomechanics model interprets those movements to calculate running metrics.
The patent describes measurements including cadence, stride length, vertical oscillation, contact time, flight time, and balance. It also discusses real-time running form analysis on a treadmill.
The process can be summarized as:
Camera footage → Body position tracking → Biomechanical analysis → Running metrics
The camera doesn't directly 'see' cadence or stride length. Instead, the system calculates those values from the movements captured in the video.
What starts as a moving image becomes measurable information about how someone runs.
3. But Can Numbers Really Tell You If You're Running Well?
Now comes a more interesting challenge.
Suppose a system measures a cadence of 170 steps per minute and a vertical oscillation of 8 centimeters. Can it immediately conclude: "Your running form scores 85 out of 100?"
Not quite.
Measuring movement is one thing. Evaluating how well someone moves requires a set of criteria. A patent from Mizuno takes an interesting approach to this problem.
 |
wipsglobal.com
JP 5314224 B1
How to score-ize a running form diagnostic system and running form
( Japanese drawing text translated into English by AI) |
The system captures a runner's movements on a treadmill and extracts biomechanical data, including joint angles and angular velocities. It then uses a scoring model developed by analyzing the relationship between these measurements and running form assessments made by experts.
Once the model has been established, it can calculate a running form score for another runner based on their movement data.
But the system doesn't stop at assigning a number. It can also provide Form Advice and Training Advice, identify areas that need improvement, and display the runner's movements alongside examples of recommended running form.
The process looks something like this:
Movement data → Biomechanical measurements → Evaluation model → Score → Coaching advice
The important distinction is that a camera doesn't inherently know what "good running form" looks like. It takes an evaluation model to turn measurements into meaningful feedback. And that's where a treadmill begins to act less like a measuring device and more like a coach.
4. A Coach That Doesn't Just Watch
What happens when a treadmill detects a problem with your running form? A patent from Zsolutionz takes the concept one step further.
 |
wipsglobal.com
WO 2015-108700 A1
Sensor-based evaluation and feedback of exercise performance
|
Its system uses camera-based sensing to observe a person exercising on a treadmill and compare their movements with reference information about proper running form. It can assess aspects such as arm angles, limb movements, head position, foot contact and torso alignment.
If the system detects an issue, it can provide corrective feedback through a display. For example, it may show the runner's avatar alongside a reference posture or visually highlight the body part that needs adjustment.
Imagine your treadmill telling you:
"You're leaning too far to the left. Try correcting your posture."
And here's where things get even more interesting. What if you don't correct your posture?
The patent describes a scenario in which the system detects running movements that suggest fatigue. If the runner fails to respond to corrective feedback, the treadmill may reduce workout intensity by adjusting conditions such as speed or incline.
In other words, the technology could follow a complete cycle:
Observe → Analyze → Evaluate → Coach → Adjust
The treadmill isn't simply collecting data anymore. It's responding to what it detects. Of course, this is a technology described in the Zsolutionz patent. It does not mean that Peloton's Tread Vision automatically adjust treadmill speed or incline based on running form. Still, it offers a fascinating glimpse into how connected exercise equipment could evolve.
When Your Treadmill Starts Watching You Run
So back to our original question: "How does a treadmill know how well I'm running?"
The answer involves much more that a camera.
Computer vision identifies body movements. Biomechanical models translate those movements into measurable running metrics. Evaluation systems interpret the data, while feedback technologies turn the results into practical coaching advice.
And as these patents illustrate, the possibilities go even further - toward exercise equipment that can respond to a runner's movements in real time.
Observe. Measure. Understand. Coach. Respond
For decades, a treadmill was simply a machine that let us run in place. Now the relationship is beginning to change. Instead of just running on a treadmill, we're entering an era in which the treadmill watches how we run.
And perhaps one day, the number we'll be most curious about after a workout won't be our speed or calories burned. It might be something much more personal:
"So, how was my running form today?"
Related patents were researched using WIPS Global. The technologies described in this article are based on the disclosures in the respective patent specifications.
< Sources & References >
Product reference, Peloton
— What's New in Treadmill Updates
Peloton Launches Foldable Peloton Tread Flex
