The primary goal of this project is to create a modular package that can be seamlessly integrated into a wide range of vehicles, enhancing their functionality and promoting the broader adoption of autonomous driving technologies. As the base of our autonomous driving packages, cars from previous MQPs were used. However, many of them had mechanical failures, which needed to be addressed. Fixing these cars was very much needed for the team for this project to be successful and efficient. The open-source Donkey Car platform was utilized as a foundation for developing the Modular Package for Autonomous Driving (mPAD) system to accomplish this objective. Donkey Car allows an RC car to control using PWM signals to send commands to the throttle and servo (steering) components. This platform collects training images alongside corresponding throttle and steering values at specific points in time, creating a comprehensive dataset for further analysis and model development. A linear regression model was created based on the collected data, which can be uploaded back to the vehicle to enable autonomous driving capabilities. Furthermore, object detection was incorporated into the system to detect stop signs and other relevant objects in the environment, enhancing the safety and reliability of the autonomous driving system. Utilizing the Donkey Car platform had some limitations, specifically in the dashboard/UI. To modernize the looks and extend the capabilities, a new UI was generated, which included the ability to graph sensor data, throttle control using a slider, and a new color palette. These changes allowed for a solution that can be reliably driven autonomously for five laps after manual driving, with object detection and the ability to collect sensor data. The modular nature of the mPAD system offers the potential for widespread integration into various vehicle types, paving the way for the future of autonomous driving.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Khung, Ian
• Colon-Heyliger, Allison
• Bleakley, Martin
• Rathi, Ishan
• Anand, Rohan

Publisher

• Worcester Polytechnic Institute

Identifier

• 107521
• E-project-050123-181235

Keyword

• Autonomous
• Machine Learning
• Robotics
• CNN Model
• Linear Regression
• Self-Driving

Advisor

• Radhakrishnan, Pradeep
• Pahlavan, Kaveh

Year

• 2023

Date created

• 2023-05-01

Resource type

• Major Qualifying Project

Major

• Computer Science
• Robotics Engineering
• Mechanical Engineering

Source

• E-project-050123-181235

Rights statement

• In Copyright

Last modified

• 2023-06-16