The objective of this project was to develope an autonomous car that drives through the simulated environment while obeying traffic laws and returns license plates and associated parking IDs using machine learning and computer vision. Linux based ROS Melodic/Gazebo was used as a simulation ground for the robot to navigate through. Google Colab was used to access python libraries such as OpenCV for computer vision and Tensorflow to design and train neural networks.
In regards to the competition the key objectives were to develop an autonomous robot which could navigate roads and detect/read license plates on parked vehicles, with obstacles including pedestrian crosswalks and other vehicles to avoid. The measures of success in the competition were plates registered and time taken to complete the course. With 6 points for outer loop plates, 8 points for inner loop plates and a cap of 4 minutes for simulation duration with points taking precedence in the final result.
Back to Portfolio