Kinarm | Interactive Robotics Brain Injury Research

Mobile nav

Task Tags Beginner

Polygon, Path, and Repeat Target

Posted on Aug 6, 2020

New in Dexterit-E 3.7, are three target types: polygon, path, and repeat. Polygons are targets that can specify up to 25 vertices. Paths are lines that connect up to 25 vertices. Repeat targets are targets that specify a circle, ellipse, rectangle, triangle, or line and draw it at up to 25 different locations.

In this task, the “Make VCODES” embedded MATLAB function block creates one example Vcode of each of the above target types to show what the new vcodes in Dexterit-E 3.7 can look like. Details on using the new Vcodes can be found in Section 10.7 VCodes – Programming Visual Stimuli of the Creating Task Programs for Dexterit-E™ 3.7 guide.

The task was compiled in Dexterit-E 3.7 and MATLAB 2013a.

Force Channel

Posted on Aug 6, 2020

Note: This task has been updated and cleaned up significantly from the first version.
Note 2: I have just updated this again so that there are different force values applied for primary vs. secondary encoder systems.
Note 3: The update I have just added allows the task to be run on the Exoskeleton (not the classic). I have also done some further clean up.

This is a sample task that implements a force channel similar to what is described here:

Scheidt et al. (2000), Persistence of Motor Adaptation During Constrained, Multi-Joint, Arm Movements. J. Neurophys., 84(2): 853-862 (https://www.physiology.org/doi/full/10.1152/jn.2000.84.2.853)

This sample is just a modified reaching task that implements a force channel between the reaching targets. The supplied protocol has 8 targets 10cm from the center. The configurable parameters are:

  • Channel width (width over which there is no force, may be zero)
  • Wall width (the width of the virtual wall that will push you back to center)
  • Wall stiffness
  • Viscosity (viscous force perpendicular to the channel, this is required to keep things stable).

Some other key concepts here are:

  • The force channel is ramped up over 1.5s once the destination target is turned on. This stops a potential violent jerk into the channel.
  • The forces are ramped down over the last 25% of the wall’s width.
  • If you break out of the channel then the forces are all off until you re-enter the channel.

This is compiled for Dexterit-E 3.6 and MATLAB 2013a.

Please note that this task is only possible to run on the Kinarm EP and Exoskeleton (not classic!) lab. The Kinarm Exoskeleton Classic Lab and Kinarm NHP Labs do not provide appropriate stiffness to make a force channel useful.

Reaching Sample

Posted on Aug 6, 2020

This is an example of a simple reaching task. This is not the same as the standard task used by Dexterit-E.

Compiled for Dexterit-E 3.8, MATLAB 2013a.

NHP Reaching Task

Posted on Aug 6, 2020

This is a sample reaching task that could be used on an NHP system.

Compiled for Dexterit-E 3.5, MATLAB 2013a.

Position controller

Posted on Jul 14, 2020

This task demonstrates how to move the subject’s arm smoothly while using the Position Controller block. It will move the subject’s arm alternatively between 4 target locations.

Forces/torques will be commanded to the Kinarm as soon as the task starts, and will continue throughout the entire task (i.e. the Position Controller is always on). There is a different sample task “Position controller enable/disable” which demonstrates how to turn the Position Controller on/off smoothly within a trial, for tasks that need to alternate between free movement and positioning the arm.

In the Trajectory subsystem/Create_Trajectory subsystem, a bell-shaped velocity profile is calculated between where the hand was at the start of the trial and the destination target. Using Stateflow, the virtual trajectory is sent point by point to the Position Controller block to create a smooth motion.

This task works on both Endpoint and Exoskeleton Kinarm labs.

Compiled for Dexterit-E 3.7, MATLAB 2013a.

Position Control – Enable/Disable

Posted on Jul 14, 2020

This task is a slightly more complex version of the position control demo. In the original position control demo the position controller comes on at the start of the task and remains on throughout. This version of the position controller demo shows how to enable and disable the position controller smoothly – i.e. no sudden jerks as the controller goes on and off. This demo is a good place to start if you plan to have a task where you drag the subject back to a starting location after they have completed a movement.

Each trial in this task starts with the position controller off (i.e. free movement). Once the initial target is reached and held for a short time the position controller is ramped on. Once the position controller is at full strength you will be moved to a destination target. After the destination is reached the position controller is ramped off.

This task is saved as MATLAB 2013a and compiled for Dexterit-E 3.6.2

Changing the Hand Feedback

Posted on Jul 14, 2020

You can change the size of the dominant hand circle by changing the value in row 1 and column 3 of the target table. The Show target block is using the columns [4 4 3 3]. This is a simple task which shows how to change the hand feedback from the default white circles.

Compiled for Dex 3.5 with MATLAB 2013a.
Technically this could be switched to [4 0 0 3] and it would look the same. If you look at the help for the show target block you can see that for a circle the values here mean: fill color column, stroke color column, stroke width column, radius column.

Background Targets

Posted on Jul 14, 2020

Normally if you would like to show 10 targets on the screen at once you would have to send 10 VCodes at a time each cycle. This is not much of a problem, but now imagine you would like to show 30 targets at a time, this can be a bit cumbersome. The “show target in background” block helps get around this. A background target is one that is always shown in its “current” state until you send a new command that overrides the current state. Background VCodes are the same as normal VCodes except for 2 small things:
The target type is (normal target type + 100). i.e. a circle is 101.
There is an ID with the VCode that can be 1-1500. This is how you interact and change your permanent target states.

This sample task makes use of the “set target in background” block. The example protocol defines 30 targets of which 10 are shown at a time. When a target is touched its color changes. This is meant purely as an example of how to properly use background targets that are defined in your target table.

The example is made with MATLAB 2013a and compiled for Dex 3.5

P.S. For some reason this task can take a very long time to compile (5+ minutes on my machine that normally takes 1-2 minutes to compile a task)

Bimanual Control Task

Posted on Jul 14, 2020

This task was developed to highlight the bimanual capabilities of the Kinarm Lab. It was built and compiled for Dexterit-E 3.5, using R2013a. It can run on either a Kinarm Exoskeleton or Kinarm End-Point lab.

Although the intent of the code was re-create the experiments performed by Diedrichsen (2007), there has not been a strict verification to ensure that all details of the experiment have been properly replicated. There are 4 task protocols included: a short demonstration of various parts of the task along with 3 protocols that implement the different experiments in the original paper.

Diedrichsen (2007). Optimal Task-Dependent Changes of Bimanual Feedback Control and Adaptation. Current Biology. 17: 1675–1679, 2007. DOI 10.1016/j.cub.2007.08.051

End-Point Perturbation

Posted on Jul 14, 2020

This task is a simple reaching task where the subject is perturbed during the reach. The perturbations will only work for a Kinarm EP, though this could easily be modified for a Kinarm Exoskeleton.

The attached model was compiled using MATLAB 2013a for Dexterit-E 3.5.