Project
| # | Title | Team Members | TA | Documents | Sponsor |
|---|---|---|---|---|---|
| 32 | Dead-Lift Assistant |
Benedict Egiebor Johan Mufuta Sean Luo |
Jonathan Hoff | design_document1.pdf design_document2.pdf design_document3.pdf final_paper1.pdf final_paper3.pdf photo1.jpg photo2.jpg photo3.png photo4.png presentation1.pptx proposal1.pdf video |
|
| # Problem Many people get injured in the gym while performing exercises due to poor technique and form. Performing deadlifts with bad form can cause serious injury. Specifically, arching your back can lead to a broken back and other spine injuries. A cheap and reliable feedback system would save a lot of people from injury. # Solution Overview Our solution for preventing injury is to use a deadlift assistant. This device that records video of someone performing a workout and provides feedback on their technique by lighting green diodes if the user performed the deadlift with proper form correctly or lightning red if they used bad form by arching their back. It uses a camera, accelerometer, proximity sensor, and computer vision to provide accurate feedback. This product is intended to be installed in gym equipment so that anyone can integrate it seamlessly into their workout. As proof of concept, we are going to specifically work on the deadlift exercise. # Solution Components ## Barbell Subsystem This will contain a PCB sensor unit that will be attached to the barbell. One of the sensors will be an accelerometer that tracks the motion of the barbell as it goes up and down to signify to the computer vision subsystem what kind of spinal motion to look for Another sensor we will use is a proximity sensor. This will indicate when the bar has started to move up as well as when the bar has reached its final position on the ground. This tells the computer vision subsystem when to start and stop polling camera data There will be LED feedback (red/green) indicating whether or not the user performed the deadlift with good form after their routine. This will be determined by feedback from the computer vision subsystem This subsystem will be controlled by a microcontroller. This will process the data from the sensors and relay them to computer vision subsystem via Bluetooth transmission This system will be powered by a 5-volt battery ## Computer Vision Subsystem A microcontroller will run a computer vision algorithm to detect bad form and provide feedback to the barbell subsystem on whether or not. To implement the algorithm, we will be using OpenCV body tracking (2) to track the angles of every major joint used in the deadlift exercise. We will write the logic behind our algorithm to determine if the user of our product is performing the exercise correctly. For example, during the upwards motion of the deadlift exercise, if the user’s back joint angles are too acute, then the algorithm will alert the user that they need to straighten their back. We are also attaching a camera module to the microprocessor to record the deadlift movement and compare the user’s form to the proper form. This will be placed at the side of the user to get a view of their body’s, specifically their backs, movement. We are having a 5 Volt battery supplies power to the accelerometer and altimeter sensors. Since they will be their own separate device/module, they will be powered by the same source. We will use a 9 Volt battery to power the microprocessor. Since the camera module is attached to the microprocessor, the camera will draw power from it, as well. Alternatively, if computer vision isn’t enough to detect the user’s spine movement, then we will attach orange ping pong balls on their back to better detect the angles. # Criterion for Success Our goal is to create a system that can successfully detect and improve a user’s deadlifting form based on user video data. For a successful project, we want the user to be able to set up the system onto the barbell and to their side before their routine. After their routine, the barbell module with use the LEDs to indicate whether the deadlift was done with proper form. Green would indicate good form while red indicates bad form. |
|||||