The Ugly Duckling

Autonomous Mars Robot to pick up targets and deliver to the mothership while avoiding obstacles. Competed at the 2019 IEEE R5 Autonomous Robotics Competition. Medium Article: https://medium.com/@rishavrajendra/autonomous-mars-rover-using-raspberry-pi-arduino-and-pi-camera-5b285be452c1

Video of the robot in action: https://www.youtube.com/watch?v=3CT76dNDTTw&feature=youtu.be

Details

3D Models

• Ugly Duckling Body
• Ugly Duckling Camera Mount
• Ugly Duckling Wing

These batteries were used to power the robot: https://www.batteryspace.com/samsung-lithium-18650-rechargeable-cell-3-7v-1500mah-5-55wh-23a-rate---inr18650-15m---un-38-3-passed.aspx

Training Data

We are releasing all the labeled images we trained the final detection model with for the robot. All the pictures were taken from a Pi Camera V2.

• Dataset

Wiring Diagram

• Wiring Diagram

MR. W Motor Controller Board by Wade Rivero

Execution

0. Compile and upload the arduino code

cd motion/
make
make upload

1. Launch the main script on the Raspberry Pi. The robot will try and navigate around the field to pick up targets and drop it in the mothership. The Tensorflow model will take around 30 seconds to boot up.

python3 main.py

Installation

Prerequisites

Tested on Python 3.5.3

pip install pySerial
pip install keyboard
pip install tensorflow
pip install numpy
pip install pandas

Follow turorials from Edje Electronics and Sendtex to train your own object detection model.

Contributors

Programming

• Rishav Rajendra - Website
• Benji Lee
• Michael Ceraso - Linkedin

Mechanical

• James Winnert - Linkedin

Motor Controller Board and Electrical

• Michael Ceraso - Linkedin
• Wade Rivero - Linkedin
• Rishav Rajendra (Explosive Electrical)

License

This project is licensed under the MIT License - see the LICENSE.md file for details