Imagine you are an automotive company having to create and execute thousands of test sequences for each step of the development stages. Even though test automation is something running in the industry for quite some time, someone still has to create test sequences (written in a programming langauge) for each test scenario (written in natural language) in order for the testing hardware to execute them.

What it does

This is were we come in to help. Let's say for example you are testing car keys and door handles. Using a LEGO MINDSTORMS robot as a testbench prototype for these tests, we will transform the test scenarios into test sequences the robot can understand and execute.

How we built it

Based on a set of training data we developed specifically for the actions the robot can execute, we will use natural language processing and machine learning algorithms to infer future actions to the robot.

This way, the time consuming task of having to build test scenarios into test sequences can be automated. The test team can now focus on the architecure of the tests and some minor corrections of what our algorithm produces.

When fully implemented, the natural language could actually become a programming language in itself.

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for TEST.pls

Built With

  • daqmx
  • labview
  • lego-mindstorms-ev3
  • nltk
  • python
  • scikitlearn
  • spacy
Share this project: