Project Logs

Collapse

• First of the new boards assembled

  George Albercook • 08/27/2018 at 02:19 • 0 comments

• New thinking on haptic actuators

  George Albercook • 08/26/2018 at 22:12 • 0 comments

  It has been very difficult to utilize the very noticeable feeling when one touches or holds an actuator.   It may be that the sensation the one can feel when reaching out to touch an actuator it from a different type of receptor.  There are 4 types of mechanoreceptors on our skin. The Meissner's corpuscles respond to light touch in the 10 -50 Hz range while Lamellar corpuscles respond to faster changes in the 250 Hz range.  Actuators are optimized for the 250 Hz range.  

  It is also interesting to learn how the body responds to multiple points of touch.  If the points are close enough the skin can't distinguish them but that was not the interesting thing to me.  If one point of contact is within the region of a particular receptors and a second point is in the adjacent receptor's region of response the signal sent indicates the union of the two regions. 

  I think this may allow the skin to have a larger dynamic range. I could not find an answer in the literature (didn't look that long) but it makes sense to me that if many regions are stimulated the the detection limit of sensation would go down.  This leads me to try a different approach.  Maybe large pads of flexible points will allow for the detection of smaller signals.  We could divide the forehead into five segments each one with an actuator.  When the actuator vibrates ever so slightly the entire section of the forehead would feel it. This may allow for finer distinction in levels and allow for better depth perception.

  Maybe they could be made out of something like this thing made to hold a bar of soap.

• v3.5 PCBs have arrived

  Nicholas Ristow • 01/19/2018 at 18:49 • 0 comments

  Recieved the first batch of PCB's for the prototype. It still uses the teensyLC, and adds a connector for a ESP8216-01 with its own linear regulator for remote telemetry. It plugs in from the back of the board.
  

  The main board also supports the i2c mux and lipo charging circuit to charge the LiPO from USB.
  

  The TOF sensor board also has a RGB addressable LED for use during testing.

  There is also a motor controller board.
  

  All of these bards have JST connector footprints, which we may or may not use.
  

• Linear Resonant Actuator tests

  George Albercook • 10/21/2017 at 01:25 • 0 comments

  They definitely start and stop quickly and it seems that you can get a more varied output but there is still a problem. It took awhile to figure out how to even explain it.  They the actuators we're on the table it was clear that the amplitude was changing with the distance to objects and when you touched the actuators you could feel the difference but when you put the device on your head they felt less lively for lack of a better term. It looks like there are two things going on here. Apparently, the driver chip has a calibration mode which we have not used yet. The documentation says they will not work comparably and properly until they are calibrated. 
  

  
  But we expect that there is another issue.  We noticed that when you reach out to touch an actuator if you press too hard you stop it from moving very much and if you touch too lightly you can't feel it move.  People appear to automatically adjust their touch pressure to maximize the movement of the actuator. Everyone seems to do it and they do it almost instantly. 
  
  Unfortunately, when the devices are strapped to your head there is no self-adjusting. We tried mounting them on foam, in foam Wells so they could move freely, stuck to the headband or stick to your head. Many of these methods worked well enough but compared to the auto-calibration that people do with their fingers they are not performing as well as they could.
  
  So our next idea is to mount each LRA on a plastic strip and adjust the length and width be of the strip until it resonates at the same frequency as the LRA. That is about 240Hz. This will give us a reproducibly secure mounting method that interferes with the actuator as little as possible. Then we can adjust how hard the actuator rests against the forehead and try different touching surfaces and shapes. Maybe a cone-shaped piece of foam that only touches at the point at low amplitude and touches a larger area at high amplitude. We will have to test and see.

  Either way ERM or LRA the system works we are just optimizing now.

• Linear Resonant Actuators

  George Albercook • 10/21/2017 at 01:06 • 1 comment

  There are two main haptic feedback technologies. Eccentric Rotating Mass (ERM) and Resonant Linear Actuators (RLA). Fancy terms for pager motors and speaker coils. Well it is not exactly that simple. The RLA have a frequency that the prefer to work at while speaker coils are designed to work at a wide range of frequencies. 

  We started with ERM actuators. They wabble around in a little circle as the motor tries to spin the off center weight. They are cheap but they are hard to control precisely. At low power the motors don't want to start so it is hard to get a large range of intensities out or them. They also take time to change power levels and tend to get warm at high power.

  So we wanted to try LRA which oscillate up and down like they are bouncing. They start, stop and change amplitude quicker than the ERM devices do and they don't heat up. So we bought some.

• Final Step

  George Albercook • 09/04/2017 at 13:55 • 0 comments

  We have the multiplexer working properly.  In the test below it is addressing each individual motor in turn and making them twitch.

• Head Mounted Version

  George Albercook • 09/04/2017 at 04:17 • 0 comments

  We made a lot of progress on a totally head mounted version. While these Sharp IR distance sensors were easy to work with and we had them on hand, there is better technology available.  

  In the past few years, the technology to measure how fast light travels  (the Time of Flight, ToF)  distancesas short as 1 millimeter to as long as 2 meters has become available in a tiny package.  

  Considering that light travels the length of a standard piece of notebook paper in one cycle of a gigahertz clock, that is kind of amazing. This technology is a form of LIDAR (Light Detection and Ranging). Think radar with lasers.  

  The photo below shows five breakout boards from Adafruit each with a VL53L0X ToF sensor made be  ST.

  
  

  Five tiny prototyping boards from Schmartboard with the DRV2606L motor drivers. 

  
  

  
  The motor driver boards attached to another Schmartboard holding an I2C multiplexer which is, in turn, connected to an Teensy LC.

  Both parts ready to be connected together.

  
  
  

• Goodbye, Electric Dreads!

  George Albercook • 09/04/2017 at 02:24 • 0 comments

  Clearly the prototype was awkward to put on and take off.  Luckily, there is nothing about the concept that requires it to keep the original configuration.  

  The very first prototype used ace bandages to hold the pager motors in place on the forearms.  We chose the forearms because the skin in relatively thin and the nerve endings are numerous and close together. Plus, the area is not "used" for much else.  

  We tried several configurations and types of fabric to make a comfortable sleeve-like configuration.
  

  Here we tried stitching them into an elastic material. 

  This turned out to be awkward to put on. When putting on a shirt, it's surprising how much it helps that the sleeves are connected to the rest of the shirt.  It's similar to fastening your own bracelet.  One arm is less usable.

  Time for a new approach. We're going to put the whole thing in a headband.

• Working with Haptic Signals

  George Albercook • 09/04/2017 at 01:56 • 0 comments

  After the rush of the Hackathon, we wanted to take a closer look at the fancy motor driver chips, which are made by Texas Instruments. 

  You might ask why not stick with the simple transistors that are working? The answer has to do with the physics of the ERM (Eccentric Rotating Mass) vibrating motors.  They have stiction. 

  When they first start moving it takes more power to start them than to keep them going.  This makes it very difficult to make them vibrate even a little. They also take time to get up to speed so it is difficult to get sharp transitions between levels. Thankfully, the DRV2605L chip addresses both of these problems and more. 

  The "L" suffix at the end of the part number indicates that it has a built-in library of vibration patterns. The library adjusts the power going into the ERM over time, resulting in refined vibration patterns. For example, it can overdrive the ERM for a short time to force it up to speed faster.

  We are still having some problems getting the DRV2605L chips to work.  At one point, we put one on the scope and it was clearly acting like it was trying to drive a huge capacitance.

  
  

  Several of the driver boards also had problems with drawing too much current.  We'll have to go through them carefully. Not at all unreasonable, considering we were working off of very little sleep and the pins of these things are half a millimeter wide. The soldering had to be done under a microscope.
  

  More updates soon!

• Real World Test #1

  George Albercook • 09/03/2017 at 23:57 • 0 comments

  One of the world's leading vision research institutions the Kellogg Eye Center  happens to be right here in Ann Arbor, MI. 

  So I called to see if I could drop in on the Blind and Low Vision Support Group meeting.  I took our electric dreadlock prototype to see is anyone wanted to try it out.  Mary was kind enough to give it a go. 

  It only took her a few minutes of using the device, before she was willing to stand up and walk around without her traditional white cane. Very exciting. 

  We are so grateful to the Kellogg Eye Center and Mary, in particular, for helping us out.

View all 13 project logs