BUZZROBOTECHS: Autonomous Box Locator Robot

Inspiration

In Amazon's vast warehouses, workers spend countless hours manually scanning boxes to locate specific items. This process is time-consuming and physically demanding. We were inspired to create an autonomous solution that could streamline this process using computer vision and robotics.

What We Built

We developed an autonomous robot that uses AprilTag detection to locate specific boxes in a warehouse environment. The system consists of:

• A Raspberry Pi with Pi Camera for AprilTag detection
• An Arduino-controlled 4-wheel robot chassis
• Custom real-time CV and depth-perception code (z-coordinate)
• Scalability in perception and navigation code

How We Built It

1. Vision System

   • Implemented AprilTag detection using OpenCV and the robotpy_apriltag library
   • Developed camera calibration for accurate distance and position measurements
   • Created robust error handling for various lighting conditions

2. Robot Control

   • Designed a simplified motor control system for precise movement
   • Implemented position-based navigation using AprilTag coordinates
   • Created a serial communication protocol between Raspberry Pi and Arduino

3. Navigation Logic

   • Developed a straightforward approach to box tracking:
     • X-coordinate determines left/right movement
     • Z-coordinate (depth) controls forward/stop behavior
   • Implemented smooth motor control for accurate positioning

Challenges Faced

1. Camera Calibration

   • Initial challenges with camera distortion and calibration
   • Solved by implementing a comprehensive calibration process

2. Motor Control

   • Complex motor control logic initially led to jerky movements
   • Simplified the control system for more reliable operation

3. Serial Communication

   • Faced issues with data transmission between Pi and Arduino
   • Implemented robust error handling and connection management

What We Learned

• The importance of simplifying complex systems for reliability
• How to effectively combine computer vision with robotics
• The value of robust error handling in autonomous systems
• The challenges of real-time motor control and navigation

Future Improvements

• Implement multi-tag detection for better navigation
• Add obstacle avoidance capabilities
• Enhance the vision system for different lighting conditions
• Develop a user interface for warehouse integration

Built With

• april-tags
• arduino
• cad
• computer-vision
• python
• raspberry-pi
• robot