Members:
Alyssa Bradshaw - akb5
Tejas Satpalkar - tejasrs2
Nathan Chung - nychung2
Introduction:
Statement of Purpose: We noticed most engineering students have bad posture. We are thinking of building a "posture corrector" which would provide feedback when a student is displaying poor posture. (possibly modeling their back posture overtime). The project itself will focus on constructing a device that uses sensors to determine if the posture of the individual is ideal, if not, the device will alert the user via haptic feedback. This can be useful for people in all white collar jobs. The idea came when Tejas had back pain while working on MPs at a desk. There are a few prototypes online, however we want to do more with the device, possibly using the data our device collects to provide useful data to the user.
Background Research:
We are interested in this project because we are interested in working with embedded systems. From our research we were able to find many variations of this project, as well as several commercial implementations. These designs either utilizing versions of flex sensors or ultrasonic/camera sensor to detect distance from the sensor to the individuals back (sensor would be mounted on the chair). Our project would be different by using a Neoprene sensor, a subset of flex sensors or LIDAR if mounted to chair. We would also connect the Arduino's used in these projects to a RPI to analyze trends.
Design Details:
a. Block Diagram
b. System Overview:
Currently there are two different design specifications that we have considered. The first has a LIDAR sensor mounted to the chair of the user, and will measure how far away the individual is from the chair. The user is responsible for "zeroing" their correct posture. The Arduino would compute if the current distance is too far away from the accepted and would trigger the haptic sensor. All of the data is fed into the RasPi which will process the data and display it on possibly a website.
The other design would involve the flex sensor array mounted on the individual's back. If the flex was too great, the Arduino would trigger the haptic sensors, and would send its data to the RPI to be displayed on the user interface.
Challenges
One challenge for the flex sensor array would be calibrating it properly, and determining how to turn the changing resistive values into data that the Arduino can handle. Another challenge will be all the coding this will involve. The data handling and exporting of data to a website is especially challenging.
Possible purchase list:
LIDAR: https://www.amazon.com/dp/B082NV4WWJ/ref=twister_B07QF236RW?_encoding=UTF8&psc=1
FLEX Sensor : https://www.amazon.com/Adafruit-Long-Flex-sensor-ADA182/dp/B01BNNNS5Q/ref=sr_1_2?keywords=flex+sensor&qid=1581733332&s=electronics&sr=1-2
Arduino uno (arduino must be able to interface to RPI
Rasberry Pi
References
», M. (2020). Neoprene Bend Sensor IMPROVED. [online] Instructables. Available at: https://www.instructables.com/id/Neoprene-Bend-Sensor-IMPROVED/ [Accessed 15 Feb. 2020].
Instructables. (2020). Posture Awareness Sensor. [online] Available at: https://www.instructables.com/id/Posture-Awareness-Sensor/ [Accessed 15 Feb. 2020].
Wareable. (2020). Sit up straight: Smart posture trainers to save your back. [online] Available at: https://www.wareable.com/wearable-tech/the-best-wearables-for-improving-your-posture [Accessed 15 Feb. 2020].
NEWWWW PLAAANNNN
Using Velostat pressure sensors to create a sensor array that will output pressure heat map to monitor.
picture from http://langster1980.blogspot.com/2016/06/designing-pressure-sensor-using-velostat.html
Will need to process a crap ton of Analog inputs if we want to measure voltage at different points.
Our project is essentially based off of https://hackaday.com/2017/10/29/hi-res-body-sized-pressure-sensor-mat/, and will try to visually map a person sitting on a chair.
The difficulty of the project will be measuring a lot of voltages at different points and the design of the sensor array. The processing of voltage values to create a map will also be difficult.
Attachments:
Comments:
|
This project would be too simple if the only way you measure the correctness of the posture is purely by the distance between an ultrasonic sensor and the user. The lidar you are using has similar functionality to an ultrasonic sensor since it does not actually output point cloud. Please think about how you analyze the sensor values. Perhaps you can have an array of ultrasonic sensors that detects different parts of the user's back? Perhaps you can generate an average distance output for each individual sensor, make it into an array of values, and compare it to the desired output? 
Posted by yuchenc2 at Feb 16, 2020 02:28
|
|
LIDAR is not appropriate for this project. As Johnny said, this requires some additional planning. You need some additional complexity, such as multiple sensors. I would recommend using pressure/weight sensors mounted to a chair. You could analyze the proportion of weight distribution across various parts of the chair, and make judgements based on that data.
Posted by fns2 at Feb 16, 2020 13:15
|
|
There are a lot of cool avenues to this project. If you wanted to make it a very complex analysis of posture I would stick to calibrating it to one person. If you want a more simple analysis, I would work on consistency between different types of individuals. My only other concern is any physical 3-D prints or wood work you may need. Make sure you know what parts you want as soon as possible.
Posted by dbycul2 at Feb 16, 2020 22:05
|
|
Cool project. LIDAR is not the best. Maybe wearing some sort of 'backpack' like device that gets the angle of your spine? If this is to be worn, maybe look into smaller alternatives to a rasbpi.
Posted by weustis2 at Feb 16, 2020 22:50
|
|
Very cool project idea! To reinforce what has been said above ^ I also agree that there needs to be more planning as far as how many sensors you plan on using and how to compare the outputs from these sensors. Weight sensors would provide more consistent data, especially if you are calibrating it to one person.
Posted by rileymb3 at Feb 18, 2020 18:26
|
|
https://hackaday.com/2017/10/29/hi-res-body-sized-pressure-sensor-mat/
Posted by tejasrs2 at Feb 20, 2020 18:31
|
|
https://nanonets.com/blog/human-pose-estimation-2d-guide/ (This way too hard but cool)
Posted by tejasrs2 at Feb 20, 2020 18:38
|
|
Here is a dump of the links that we went over: https://www.tekscan.com/force-sensors https://www.adafruit.com/product/1361 Look for TNCduino or other high input arduino
Posted by dbycul2 at Feb 20, 2020 18:54
|
|
Approved, Feb 20th. Thanks for talking with course staff for ideas and design input.
Posted by fns2 at Feb 20, 2020 18:56
|