Forum Discussion
VR Gloves (my project)
Hey everyone, I know I've had my vr glove project under my projects for a while, but recently I've been having more free time and actually got started working on it the last week or so. The idea is...
There are many issues with optical systems. While they've gotten decent, you need only look at the leap motion to see the problems with it.
The biggest issue being that optics require line of sight. If you stick your had behind your back, it can't track it unless you have a multiple camera setup. While a multiple camera setup can help, if you turn your hand sideways or upside down, it also causes issues. Sideways, it can't determine which finger is the one moving, and upside down can sometimes be registered as the other hand. There's no inherent way to fix this issue.
As well, multi camera setups ( like those used in motion capture ) require a lot of room, and as people that use the kinect have shown, it's not really worth wile for them to set up the space to use the thing, let alone having to run all the wires and such for multiple cameras.
At the moment, speed can also cause problems. While there's less lag in something like the leap motion, the leap motion also has a very limited field/range. And while the Kinect has a larger field/range, it's limited in it's accuracy, not that the leap motion is super uber accurate anyway. It only gains it's accuracy due to having the much more limited field/range.
My point is, I don't see optics taking off as a VR viable alternative to something like a glove, which would be accurate much more often then not compared to an optical alternative.
A LITTLE UPDATE:
So, I may have mentioned the whole thing about having limited analog inputs, not being able to find a decent gryo locally, etc etc. Well, I did some looking, and it turns out there is a spiffy circuit called the MPU-6050. This spiffy little chip has both an accelerometer and a gyro in it. This is awesome, cause that's what I needed really. I was effectively going to make this chip out of separate pieces of circuitry. The only problem is, it's a surface mount chip. Tiny one at that.
So, I looked around a little more and found people are making breakouts for it. :D Also, I found some in china for $3.05. So... I bought two. Unfortunately, it's coming from china, so it can take 3-4 weeks, sometimes longer (generally around 3 weeks though) BUT, that also means price reduction in required parts. Instead of being around $20 per chip ($80 for all accelerometers and gyros for one arm) it's $6.10. Considerably cheaper. :D Beside that, this MPU (Motion Processing Unit is what it is I believe) uses an I2C system. That means it only takes 2 wires to talk with it. Past that, you can daisy chain them. So instead of needing 12 wires to talk with all the sensors, I only need 3. (2 wires for talking, one wire for setting the address of the second chip. This third wire is a simple on/off, so it doesn't take any of the analog pins and only requires a digital pin)
This is super useful as it lets me have 10 more possible analog inputs. Still not quite where I need it for the second prototype (as that needs roughly 19 potentiometers... so 22 analog inputs total. When I get to this point, I'll probably use a 16 channel multiplexer, maybe two 8 channels {for speed purposes. You have to wait a few ms between switching channels so that they don't bleed into each other, and doing it parallel would thus make each cycle faster})
A lot of techno jargen, I know. But maybe I explained what I'm talking about good enough?
EDIT:
As toward the cost, currently, my setup would be over $100 per arm. With this new cheap chip (around $4.50 in 5000 unit bulk just for the chip from digikey {boards I bought are cheaper cause they are direct from china}, a custom arduino would be around $4 in cost for all parts, and the custom pcb would be around $6 a piece, and 19 potentiometers/trimmers at around $3 total) the price could theoretically make the cost of materials be around $20 per glove instead of $100. (if stuff is bought in bulk. No promises on all that though, obviously.
I do want to work on this more. T_T but I have to wait for the chips from china to come in.
The biggest issue being that optics require line of sight. If you stick your had behind your back, it can't track it unless you have a multiple camera setup. While a multiple camera setup can help, if you turn your hand sideways or upside down, it also causes issues. Sideways, it can't determine which finger is the one moving, and upside down can sometimes be registered as the other hand. There's no inherent way to fix this issue.
As well, multi camera setups ( like those used in motion capture ) require a lot of room, and as people that use the kinect have shown, it's not really worth wile for them to set up the space to use the thing, let alone having to run all the wires and such for multiple cameras.
At the moment, speed can also cause problems. While there's less lag in something like the leap motion, the leap motion also has a very limited field/range. And while the Kinect has a larger field/range, it's limited in it's accuracy, not that the leap motion is super uber accurate anyway. It only gains it's accuracy due to having the much more limited field/range.
My point is, I don't see optics taking off as a VR viable alternative to something like a glove, which would be accurate much more often then not compared to an optical alternative.
A LITTLE UPDATE:
So, I may have mentioned the whole thing about having limited analog inputs, not being able to find a decent gryo locally, etc etc. Well, I did some looking, and it turns out there is a spiffy circuit called the MPU-6050. This spiffy little chip has both an accelerometer and a gyro in it. This is awesome, cause that's what I needed really. I was effectively going to make this chip out of separate pieces of circuitry. The only problem is, it's a surface mount chip. Tiny one at that.
So, I looked around a little more and found people are making breakouts for it. :D Also, I found some in china for $3.05. So... I bought two. Unfortunately, it's coming from china, so it can take 3-4 weeks, sometimes longer (generally around 3 weeks though) BUT, that also means price reduction in required parts. Instead of being around $20 per chip ($80 for all accelerometers and gyros for one arm) it's $6.10. Considerably cheaper. :D Beside that, this MPU (Motion Processing Unit is what it is I believe) uses an I2C system. That means it only takes 2 wires to talk with it. Past that, you can daisy chain them. So instead of needing 12 wires to talk with all the sensors, I only need 3. (2 wires for talking, one wire for setting the address of the second chip. This third wire is a simple on/off, so it doesn't take any of the analog pins and only requires a digital pin)
This is super useful as it lets me have 10 more possible analog inputs. Still not quite where I need it for the second prototype (as that needs roughly 19 potentiometers... so 22 analog inputs total. When I get to this point, I'll probably use a 16 channel multiplexer, maybe two 8 channels {for speed purposes. You have to wait a few ms between switching channels so that they don't bleed into each other, and doing it parallel would thus make each cycle faster})
A lot of techno jargen, I know. But maybe I explained what I'm talking about good enough?
EDIT:
As toward the cost, currently, my setup would be over $100 per arm. With this new cheap chip (around $4.50 in 5000 unit bulk just for the chip from digikey {boards I bought are cheaper cause they are direct from china}, a custom arduino would be around $4 in cost for all parts, and the custom pcb would be around $6 a piece, and 19 potentiometers/trimmers at around $3 total) the price could theoretically make the cost of materials be around $20 per glove instead of $100. (if stuff is bought in bulk. No promises on all that though, obviously.
I do want to work on this more. T_T but I have to wait for the chips from china to come in.