Autonomous Ticketing Agent

Team 7 aka MACH7

MAE 148 Final Project SU24

Table of Contents

1. Team Members
2. Abstract
3. What We Promised
4. Accomplishments
5. Demonstration
6. Challenges
7. Robot Design
8. Electrical Diagram
9. References
10. Contacts

Team Members

• Aryan Palaskar - Computer Science
• A.J Olivares - Cognitive Science: Machine Learning
• Edward Lee - Mechanical Engineering

Abstract

The goal of our project, the "Autonomous Ticketing Agent", focuses on implementing an automatic license plate recognition (ALPR) system along with the vehicle control system within ROS2. We aim to do this so that we could automously scan for license plates in parking lots and ticket any vehicles that do not have a valid permit. By leveraging the capabilities of ROS2, we aim to design a robust framework that ensures an accurate license plate detection and effective autonomous control. This innovative approach not only enhances the efficiency of traffic management, but also contributes to the development of smart parking lots.

What We Promised

Must Have:

• A working license plate detection model for the camera that can recognize and store the individual letters/numbers.
• A script that compares the license plate numbers with a list of valid permits.

Nice to Have:

• An automated route for the robot to follow using the GPS.
• An obstacle avoidance model using the LiDAR.

Accomplishments

• We utilized the OAKD camera through the use of a license plate recognition CV model that was developed and posted on Roboflow.
  • We were able to detect the individual numbers/letters on the license plates and store them with about a 90% accuracy.
• We created a script that compares the license plates with a database of license plates that have proper permits.
  • It can then alert the user if the license plate does not match or if it’s valid.
• We also created a quick GPS route that can work alongside the license plate detection system.
  • However we didn't have enough time to implement it into ROS2, so we had to run it on DonkeyCar.

Demonstration

IMG_1371.1.1.mp4

Recording.2024-09-06.163432.1.mp4

Challenges

• There were issues on how to get the license plate recognition model and the license plate comparison script onto ROS2.
  • The solution was to just create a ROS2 node that launches alongside the other ROS2 nodes, allowing us to just import our license plates script directly onto ROS2.
• There were also issues on getting the GPS to stay on its path accurately.
  • We were previously inside of the EBU2 courtyard, so the solution was to go away from any buildings in order to prevent the GPS signals from bouncing off the walls.

Robot Design

Electrical Diagram

References

• Open ALPR
• Roboflow License Plate Vision Model

Contacts

• Aryan Palaskar - [email protected]
• A.J Olivares - [email protected]
• Edward Lee - [email protected]