CMPS 261 Data Visualization Winter 2019

Dongshuo Li, University of California Santa Cruz, Computer Engineering

Visualization of a Somatosensitive Actuator

About the Project:

The Electrical and Computer Engineering department in University of California Santa Cruz has developed a soft somato-sensitive actuator (Picture Up). The actuator is in a shape of a finger which is driven by the compressed air to bend or flick. To design a computer algorithm to control the actuator, a math model between the motion tracks of the actuator and the input pressure is necessary. An Effective data visualization method is essential to obtain the math model. To achieve the data visualization, I explore to put some reference points on the actuator, use the OpenCV library via python to do the motion tracking of a testing actuator and extract the graph coordinates. The matlab toolbox can provide me a way of plotting the graph via the graph coordinates.

The Project is divided into three parts, actuator creation; motion Tracking and data visualization.

Finger Creation:

The Machine Interaction Lab in University of California Santa Cruz has successfully developed molds for the Soft-Robotic-Gripper. The matertial to pour in the molds is the Dragon Skin 30. The whole process includes mixing the dragon skin in the mold, vacuuming the mold and bake the whole part in the oven at 60 Celsius degree for 10 minutes.

Motion Tracking:

OpenCV and python language is a good choice to achieve the motion tracking part. To run the code, Jupyter notebook is one of the good platform. It runs on a browser without considering building enviornment. All the software and library here are open-source. With both software installing, a video of the an actuator sticking some reference points is recorded as an input. A code that is written to by Python using OpenCV to run on Jupyter notebook has the feature to detect the reference point and extract the pixel coordinates of the center of the reference point as a csv file. For more information click Motion-Tracking-code OpenCV Jupyter Notebook.

Data Visualization

Matlab is one of the powerful tool to sort the data from csv file generated by OpenCV and do the data visualization. The whole script has two major parts.

Correctly sort the data because the code from previous part does not guarantee the correct point order in each frame

plot the animated motion traces of three reference points

The data visilizaiton code is here and should run in matlab. Data-Visualization

Results of the Project

The videos downwards are the input video and motion traces.

User mannual

Download the video of actuator movement with reference point on that

Download the Jupyter Notebook and follow the instrutions on their website to install it

Download the OpenCV and follow the instrutions on their website to install it

If last two procedures work perfectly, you should run "Hello World" of Python on Jupyter Notebook; if not, you need to google it until it works

Download the Motion-Tracking-code and run it on Jupyter Notebook, make sure the input path is correctly linked to the video

If last procedure works perfectly, you should get a video file as the one in motion tracking part and a csv file called center point; if not, check video input path, try not to change the rest of the code

Download the Data-Visualization to run it in matlab, make sure the input csv file is correctly linked

You can also download the paper here here.

Acknowledgement

Great thanks to Max Belman and Jacqueline Clow in University of California Santa Cruz for their help on the motion tracking code compliation. Great thanks to Colin Martin for his help of video capturing of the actuator.