Github

VisionBot is a collection of several utilities for obstacle detection algorithm path-finding. The project contains the following modules:

  1. Optical flow calculator: Calculate the optical flow to a given video or a video stream and plot the flow vectors on the top of the video.
  2. MagicMotion: A video clip generator for optical flow verification. The module applies translation, rotation or stretching to a static image.
  3. IQIF motion detector: preliminary results to the obstacle avoidance algorithm.

Tech Stack

  • Development: The motion detector demo is implemented on a Raspberry Pi.
  • Programming languages: The major part of the application is written in Python and Qt, and the IQIF neural simulator is written in C++ in consideration of efficiency.
  • Algorithms: Most of the image processing algorithms are called from the OpenCV library with Python binding.

Results

Motion detector real-time demo.

Acknowledgements

  1. Thanks to my teammate Cheng-Fu Yeh‘s algorithmic insights to the obstacle avoidance module and he is also the author of the IQIF (Integer Quadratic Integrate-and-Fire) neural simulator.
  2. Thanks to the team leader Cheng-Te Wang for his precious suggestions for the project.