Multifunctional Robot Design (UESTC-UoGJEP Creative Lab)
A smart car for tracking, crossing bridge, and transporting a ball to a target location by Autonomous Control Design and ML Coding (2022)
Project Overview
This was a one-year group project in the UESTC-UoG Joint Educational Program Lab to develop a multifunctional smart car capable of line tracking, bridge crossing, and transporting a ball to a target location through autonomous control and computer vision.
I was primarily responsible for the computer vision and tracking module, using an OpenMV camera and Canny edge detection to estimate the road direction and support autonomous steering.
Role
Computer Vision Developer
Tracking Algorithm Design
Embedded Implementation
System
Autonomous Smart Car
OpenMV Camera
Vision-based Navigation
Skills
OpenMV
C Programming
Computer Vision
Canny Edge Detection
Computer Vision Tracking
Our tracking algorithm was based on Canny edge detection and window-based image analysis.
The lower region of each image captured by the OpenMV camera was divided into multiple windows of equal area. For each window, the standard deviation of pixel values was calculated to distinguish the road surface from the surrounding environment.
Regions identified as background or surroundings were discarded, while regions classified as part of the road were retained. The center road region was then selected to estimate the desired travel direction.
Based on the estimated road position, a direction line was generated and used by the autonomous controller to determine whether the smart car should turn left, turn right, or continue straight. The position of each selected window also corresponded to an approximate steering angle.
System and Tracking Results
