BOX
Purpose
Box is a project to work on different control and sensing methods while also providing a means to get research data on. The end goal is a roughly box-shaped container with an attachment point that moves in the X/Y/Z axes (no pitch/roll/yaw). The attachment point will allow a user to pick and place items, real or virtual.
The Goal Experiment
Using the BOX, users will be asked to pick and place objects within the range of motion. Set A, the non-treatment option, will use the box in a non-interfering mode. Set B, the treatment option, will have users complete the same task, quasi-virtually. The BOX will provide the same forces, without an actual object present. Finally, Set C will have the user complete the same virtual task as Set B, but while wearing a VR headset that will show a virtual representation of the object to be moved.
Users will have their muscle use monitored by EMG sensors placed on the forearm, bicep, tricep, and deltoid. Their time to complete the task and the path of motion will also be logged. Finally, their opinions will be recorded during an exit interview.
The goal of the experiment is to show a less than 5% difference in muscle activation, time to complete the task, and path of motion.
01 May 21: Rack and Pinion is not the way
The goal was to create a means to measure the true position on the rail to compare to the cable-measured position. This is a classic implementation of the Series Elastic Actuator method, where an elastic component within the system causes the position calculated at the distant end to be different from the true position. This allows for some "slop" in movement but can spell trouble if it is not properly handled.
To fix this I considered multiple solutions:
A laser measuring system
A spring-loaded string attached to a potentiometer
A series of buttons that would be tapped by the wheel system
A continuous force sensor
An optical sensor reading banded codes in the V-slot
A geared belt (ala 3D printer)
But I ultimately settled on a rack and pinion system attached to a rotation sensor. The thought was it would be cheap enough to prototype and I could 3D print essentially the whole system. Since delays kill projects, that's what I went with.
Unfortunately, the end result was less than stellar and revealed some issues that the CAD model did not. Chief among them was skipping. When the pinion was not perfectly aligned, it would entirely miss the rack, thus missing counts and ruining the whole purpose of the thing. This could be potentially remedied by tightening the connection, but then the system became harder to move. This is supposed to be as frictionless as possible, so that won't do.
There are some other ideas from the list that I don't hate, but they're for the most part not rapidly iterable. For now I'll think on it and come up with a solution.