6
\$\begingroup\$

I would like to track the movement of a persons hands and feet in relation to their waist. I imagine them wearing something around their waist that can measure how far away a sensor attached to each hand and foot is.

What kind of sensor would I be looking for to do this? Everything needs to be attached to the person so can't be a done via a stationary camera.

I thought IR or some kind of sound sensor but would those work when not in line of sight such as when they are walking/running and their feet are behind the front of their waist?

\$\endgroup\$
5
\$\begingroup\$

That's a very interesting problem!

I think it could be done with transmitting and receiving transducers on a belt. An ultrasonic pulse, generated by an MCU, which resets a timer, is transmitted. An MCU with receiving and transmitting transducers, on, say, the right wrist, detects the pulse, and immediately retransmits it. The belt MCU detects the retransmitted pulse, and stops the timer. From the speed of sound, and the time for the outward and return journey, the total distance can be calculated. Halving the distance gives the distance from the wrist to the waist.

The distances are relatively small, so enough signal should be present, even if the transmitter and receiver are not line-of-sight. Amplification will be needed for reception.

I'll leave you to work out the details for all four wrists and ankles. :) Some form of encoding will be needed so that the four remote units can be distinguished from each other.

A Google search for similar techniques turned up this, which uses a passive reflector on the hand, with the transmitter and receiver on a microphone stand. I can't see that simpler technique working for Glen's application.

\$\endgroup\$
0
\$\begingroup\$

The distance from a person's foot to their waist won't change much while they're walking (as opposed to, say, standing up from sitting). Except when they bend their knee, their foot will always be about a leg's length away from their waist. If you're using a distance sensor, it will need to be further from the leg's center of rotation (hip). To look at it another way, the foot behind that you're stepping off of, and the foot ahead that you're stepping on to are about the same distance from your hip.

To be totally mobile (without protruding booms to aid distance measurement, f/ex), you'll probably need angle sensors at each joint between your reference point (the person's CofG, f/ex) and the body part you want to locate. Some joints will need 2 angles to describe them; the shoulder and wrist would need 3.

You might use more sensors in different combinations to reduce ambiguities or improve accuracy, or to reduce the math if you're tracking the parts in real-time. Eg., if I reach over my head, the distance from my waist to my hand has most of the position information, but distinguishing between my hand on either shoulder wouldn't be helped at all by its distance from a sensor in front or in back of my waist.

\$\endgroup\$
0
\$\begingroup\$

Use 3 axis accelerometers.

Put two on the torso ( top, bottom) Two on lower leg ( one at knee, one at ankle) Two on forearm ( one at elbow,one at wrist)

That gets you quaternions for the ends of the limbs, and it all acts like a Wii controller, only more accurate as the "bar" is longer.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.