2022SummerTeam4

From MAE/ECE 148 - Introduction to Autonomous Vehicles
Jump to navigation Jump to search

Team Members

Aditya Tomar (COGS)

Steve Morales (CE)

Minh Quan Ly (MAE)

Car Assembly

Final Assembly

ISO VIEW TOP VIEW ASSEMBLY WITH LIGHT

Mechanical Design

All electronic hardware is mounted to the chassis with 3D Printed Parts and machine screws. One central "bridge" spans the length of the chassis and hardware has adapters that are designed to snap into place. The "bridge" has indexing tabs so that hardware can be mounted at specified locations. Hardware components are fixed to the adapters using machine screws. The adapters then snap into the "bridge" at any of the indexed slots. This tool-less design allows for flexibility as parts can be easily and quickly moved without the need for screwdrivers/allen keys.

BRIDGE GO PRO MOUNT VESC MOUNT

JETSON CASE JETSON MOUNT SWITCH MOUNT

Wiring Schematic

Wiring Schematic


3 Autonomous Laps

Donkey

Robocar ROS2

Final Project

Proposal

1. Create a Dynamic Region of Interest (ROI) that changes depending on a variable, such as Track Curvature and Error from the Centerline.

2. Create a Canny Filter for Lane Detection as alternative to Color Filter.

Overview

We wanted to create an alternative method for lane tracking using Canny filter and Hough lines for our final project. We created an extended camera mount to give our program more information to create the tracking lines. Our program contains a 2 windows:

  • a window that contains 4 different modes to view an image accessible through key presses
  • a window that has sliders that can change parameters for the edge detection in real-time:
Different Modes:

1. Frame-within-Frame: ROI w/ Canny Edge

2. Cropped ROI in Grayscale

3. Cropped ROI w/ Canny edge

4. Normal Frame with Hough lines for lane tracking

In the program, we allowed the user to move between these modes with keys (left: <-, a; right: ->, d)

Achieved

1. User-Defined ROI

2. Canny Filter with sliders to adjust filter settings

3. Hough Lines for Lane Tracking

4. Confirmed algorithm works on Jetson Nano

Demonstrations

Live Demo

Live Demo on Jetson

With More Time

1. Create ROS2 Package and Integrate with Dominic’s ROS2 Docker Container

2. Output Error for Navigation with Lane Detection

3. GPU Acceleration

GitHub

https://github.com/adityatom19/robocar

Presentation

https://docs.google.com/presentation/d/e/2PACX-1vSC485RczUsAIokHKKcXdEPzcMO56Y6pm2nkpOsW2YOWzCOyVQX2kH_qLydGOATSprEIcpT4m3tNX4W/pub?start=false&loop=false&delayms=3000

Acknowledgements

Ivan Ferrier

Dominic Nightingale

Mark & Alexis from ECE Makerspace

Jack Silberman