jaku-jaku/autonomous-maze-solver-stewart-platform

MTE 380 Course Project

Proj_StewartPlatform

UW MTE 380 Course Project - 2021 Tron Authors: Jack, Jeremy, Tsugumi, Zong

Summary

Autonomous maze solver with Stewart Platform via a single WebCam.

Program Directory: ./Path_Finding

Possible Command line inputs:

• For local image processing (demo) based on camera feeds from last successful detection

  python3 main.py -m static

• For live image processing with live camera feeds

  python3 main.py -m run

• For hsv calibration

  python3 main.py -m calib python3 main.py -m calib -c live

• Optional* specifying usb.serial port:

  python3 main.py -m run -p /dev/tty.usbmodem146300

CV Outputs:

CAD Renders:

List of Items

Hardware List:

• Servos x6
• MCU board
• WebCam: Logitech C270

Repo Contains:

• Low level C code
• Inverse Kinematics Python
• Full integration of the solution
  • CV for maze detection and extraction
  • Framework for deploying algorithms
  • CV for marker detection and extraction
  • Grid Occupancy for binarizing maze
  • A star algorithm (using library)
  • Custom A* algorithm (Currently Used), with energy/heat map to make the path more centered and optimized
  • Control Algorithm and integration pipelines to hardware interface

System Sketch

Flow Chart of the completed pipeline:

Actual Photo of the Platform

External_API

• http://kieranwynn.github.io/pyquaternion/