Human Motion Differentiation with 6DOF IMU

Projects Discussion and Showcase Science and Measurement
1

Hi all,

I will be developing a project of differentiating basic human motions with MPU6050 as my undergrad thesis.
The basic idea of the system is to differentiate simple human motions such as walking, jumping, incline/decline walking cycling and etc. The sensors will be attached to user's ankle and the system should be able to send results to an Android device via bluetooth.

The general idea of the system follows:
MPU6050 --> Kalman Filter --> Signal Recognition with Arduino --> Bluetooth --> Android Device.

I am having trouble with figuring out how to process the sensor data into a useful algorithm that recognize the motion performed.
I understand i should collect the data empirically on each motion performed first, and try to develop the algorithms/equations. Is there any good methods that i could refer to? So far i only found a useful link describing how to count steps:

and modifying code from forum user Lauszus to get the first two part done:
http://forum.arduino.cc/index.php/topic,58048.0.html

The bluetooth module used will be a cheap HC-05.

Anyone who could give me any kind of advice on how the project should go and how to get results will be highly appreciated. :slight_smile:

Thank you for any reply in advance. XD

Terrence.

2

Nobody has replied to this, so I will.

There are plenty of methods to calculate the orientation using the MPU 6050. In three dimensions, the quaternion method is probably the easiest method to apply on arduino. It's not simple, but there are plenty of examples. You can write your own, or try and adapt someone else's solution.

A lot of the methods people describe only solve in 1 dimension, effectively they solve the simplified segway problem. Quaternions are not useful for that, but for the 3-d problem, they are.

3

If you don't know about Seb Madgwick's human gait-tracking work, take a look at this site. He offers his own IMUs and code (open source) Gait tracking with x-IMU – x-io Technologies

4

jremington:
If you don't know about Seb Madgwick's human gait-tracking work, take a look at this site. He offers his own IMUs and code (open source) Gait tracking with x-IMU – x-io Technologies

Thank you this is really helpful.