Any idea about how determine body movement upun the accelerations?

  • Context: Graduate 
  • Thread starter Thread starter niko2000
  • Start date Start date
  • Tags Tags
    Body Idea Movement
Click For Summary
SUMMARY

This discussion focuses on developing an algorithm to determine a person's activity based on acceleration data collected from accelerometer sensors on mobile phones. The primary goal is to create software that can automatically detect falls and call for emergency assistance, particularly for elderly individuals. The user has experimented with Fourier transforms but found them insufficient for accurate detection. A suggested approach involves treating the x, y, and z components of acceleration separately and utilizing numerical integration techniques, specifically the Runge-Kutta 4 (RK4) method, to analyze movement patterns effectively.

PREREQUISITES
  • Understanding of accelerometer sensor data and its applications
  • Familiarity with numerical integration techniques, particularly Runge-Kutta 4 (RK4)
  • Basic knowledge of Fourier transforms and their limitations in motion detection
  • Experience with data analysis and pattern recognition in time-series data
NEXT STEPS
  • Research the implementation of Runge-Kutta 4 (RK4) for numerical integration of acceleration data
  • Explore machine learning techniques for activity recognition from sensor data
  • Investigate advanced signal processing methods beyond Fourier transforms for motion analysis
  • Study existing algorithms for fall detection and their effectiveness in real-world applications
USEFUL FOR

This discussion is beneficial for software developers, data scientists, and researchers focused on wearable technology, health monitoring systems, and motion analysis algorithms.

niko2000
Messages
50
Reaction score
0
Any idea about how determine body movement upon the accelerations?

Hi,

I am trying to implement an algorithm which would determine the person's activity upon the received acceleration measurement data.

I am measuring the accelerations using accelerometer sensors on mobile phones and send the data to the server where I process the measurement and try to determine the type of person's movement. The key idea is to develop the software which would automatically call the emergency if the person would fall (useful for old people).

So far I was trying to use Fourier transform and measure the frequency components but this by itself is not accurate enough. Enclosed I am sending an example chart of walk.

In case of fall it is easy to guess what happened by watching the chart but the goal is to implement an automatic detection.

I would really appreciate if you could share your idea about analyzing the patterns and determine what happened. It would be great if a computer could determine for example if the person is walking, running or if a person fell.

Thanks for help!

Niko
 

Attachments

  • walk.jpg
    walk.jpg
    33.2 KB · Views: 467
Last edited:
Physics news on Phys.org
First, the good news- you can treat the x, y, and z components separately.

Now, the bad news- if your acceleration data is not constant, you will need to do a numerical integration of the data.

Given a(t) as the acceleration for a particular component, v(t)= \int_{T_0}^T a(t)dt+ v_0) is the speed at time T in that direction assuming speed v0 at time T0 in that direction.

p(t)= \int_{T_0}^T v(t)dt+ p_0 is the component of position itself at time T given component p0 at time T0.
 
The bad news is hardly bad :) Numerical integration of data is relatively easy. For a relatively accurate and stable numerical integrator, look up Runge-Kutta 4 (RK4) which is a forth-order numerical integrator
 
How can you determine the type of motion by integrating the graph?

The goal is to determine the type of movement by analizing graph - whether it is walk, run, fall...

I wrote about the concept here: http://mypetprojects.blogspot.com/
 

Similar threads

  • · Replies 33 ·
2
Replies
33
Views
4K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 10 ·
Replies
10
Views
2K
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 78 ·
3
Replies
78
Views
7K
  • · Replies 7 ·
Replies
7
Views
2K