Use Codecraft 2.5 New Interactive Lessons with M.A.R.K.

In this article we will show how to use one of the Codecraft 2.5 new features, namely interactive lessons, that blend seamlessly graphical programming environment with step-by-step tutorials and valuable knowledge.

To get started, go to , Codecraft’s main page, where you will see a great range of lessons and courses you can try. If you’re using MARK for the first time with this computer, follow the steps in First Use Guide with Codecraft 2.5 video.

After the drivers are installed, and you have the latest firmware, you can pick Autonomous Driving Course Materials of the Artificial Intelligence Series, a middle school level course on AI with M.A.R.K from Codecrafts’s main page. Normally when you’re going through the course by yourself or teaching it to students, you would start from the first lesson. However, in this article to try more advanced functions of MARK we’ll skip straight to lesson 10, Using Computer Vision Sensor for Object Detection.

Go through the steps one by one and complete two basic tasks: Animal recognition and Traffic Signs Detection. After basic tasks are done there is a Work Display part, where in class setting you can ask the students to share the results of their work with classmates. And finally if you feel that basic tasks are not challenging enough (they grow in difficulty as course advances though), there is an Expansion task section, where the students need to implement a more advanced program based on what they’ve learned in this and previous lessons.

And of course, beyond the course content there is a lot of options to extend the capabilities of MARK, starting from adding additional Grove Modules to training your own custom models. Let’s have a look at how you might choose to expand on the content of the lesson 10 in classroom setting. After the students already have developed understanding on how to use pre-trained models included with MARK, we can teach them to perform object detection and image recognition with custom user-trained models. For that, go through the Colab Notebook we prepared, which includes 5 steps, that are present in many Deep Learning task workflows:

  • Data collection
  • Data preprocessing
  • Training
  • Improvement
  • Model deployment to production

After you have a trained model, copy it to SD card or (advanced user option)write directly to device’s flash memory using kflash_gui tool. Either way, next step is to define custom model in Codecraft.

Click on screenshot to see the full picture. The category names are from labels.txt

After custom model is defined, two new blocks will appear in Codecraft blocks tab. Let’s write simple code that makes MARK wander around until it encounters and obstacle. Then it performs image classification on the obstacle and if it belongs to “devils_ivy” class, turns on green light and takes a picture.

Stay tuned for more articles and updates on MARK Kickstarter campaign.

For more information on MARK, Grove Zero series, Codecraft and other hardware for makers and STEM educators, visit our website,

About Author


April 2020