Active Driving AI Assistant

WINLAB Summer Internship 2022

Group Members: Kishan Patel, Lucas Cordova, & Atharva Belsare

Project Objective

This project study will develop a naturalistic driving monitoring and intervention system. Our interactive system will use multiple in-cabin vehicle sensors to assess driving conditions and performance in real-time and draw a driver’s attention using a voice-based interface to improve their performance. We will use AI-driven techniques to continuously assess the effectiveness of our interventions and make adjustments to maximize driving safety and quality.

Reading material:

• ​Tong Wu, Nikolas Martelaro, Simon Stent, Jorge Ortiz, and Wendy Ju. 2021. Learning When Agents Can Talk to Drivers Using the INAGT Dataset and Multisensor Fusion. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 5, 3, Article 133 (Sept 2021), 28 pages.
• ​Tong Wu, Enna Sachdeva, Kumar Akash, Xingwei Wu, Teruhisa Misu, and Jorge Ortiz Toward an Adaptive Situational Awareness Support System for Urban Driving.

Week 1 Activities

Download videos of car approaching stop sign (Youtube). Learn Video Processing Using Python (cutting, concentration, noise cancellation etc). Work with ROS and Python to implement a program to estimate the total stopping distance of a vehicle based on its speed. Familiarize with Orbit/Cosmos testbed and procedures. Set up development environment and necessary tools.

Week 2 Activities

Look into achieving a program to calculate distance of car to stop sign at the parking lot outside. Our project will be the first step in a major multi-step project that will allow a car to operate autonomously. Familiarize ourselves with NJ MVC Driving Laws. Find related studies and Youtube videos relating to AI detection. Begin researching on object and motion detection techniques.

Week 3 Activities

Implement basic object detection algorithm using OpenCV and Python. Implement the videos to YOLO (Real time Object Detection). Collect more VIDEOS and compare the theoretical results to experimental results. identify the suitable camera.

Week 4 Activities

Research related Datasets and codes, and create our own dataset. Choose images and check if its correct. Install the required packages & run the code in Python Integrated Environment. Learn image Processing.

Week 5 Activities

Build upon the object detection algorithm from the previous week and develop a motion detection algorithm. Test and evaluate the performance of the motion detection algorithm on various videos. Optimize the algorithm for maximum accuracy and speed.

Week 6 Activities

Implement the AI driver assistant module to promote driver safety. Integrate the object and motion detection algorithms with the AI driver assistant module. Test and evaluate the performance of the AI driver assistant module

Week 7 Activities

Research and implement ways to improve the accuracy and reliability of the AI driver assistant. Test and evaluate the performance improvements made to the AI driver assistant.

Week 8 Activities

Begin working on integrating the AI driver assistant with the ROS framework. Test and debug the integration of the AI driver assistant with ROS.

Week 9 Activities

Complete the integration of the AI driver assistant with ROS. Test and evaluate the performance of the AI driver assistant within the ROS framework.

Week 10 Activities

Conduct final tests and evaluations of the AI driver assistant system. Prepare a poster board that summarized the methodology, results, and applications of the project. Document and present the results and findings of the internship project to the team.