Aims and goals - prototyping small scale autonomous modular robot platform. Which is inexpensive, using open source hardware and software, easy to modify and upgrade. The challange for us is to use computer vision technologies to analyze and evaluate minerals from the Moon. After the colected data is processed, to make decision whether the target is worthy being obtained.

Solution - our proposal for completing this challange is constructing small scale robot platform using mini computer Raspberry Pi 3, and computer vision system developed on Python and autonomous GPS positioning system. We can use arteficial inteligence based computer vision using Tensor flow algorithm. The technology behind this is using predefined models based on neural networks, and compare them to the colected video stream from camera. The robot should be driven absolutely autonomous.

Realization - since this is a protype robot we have developed much simpler computer vision, which in the case detects colors and shapes. Based on that it can make decision wether to mark the detected target, using a sprinkler system, or to skip it. The autonomous control however is developed completely and is working without issues. This is achieved using Pixhawk controler in which you can define GPS coordinates and make autonomous path which the robots follows. The controler also provides alternative control method which uses FrSky control unit. The robot is fully constructed using additive technologies such as 3D printing with PLA plastic. The parts for the construction can be easily multiplied and modified depending on the needs.

Source code for the computer vision of the Moon rover: https://github.com/doggo46/doggo46/blob/master/Com...