Jigsaw: The Puzzle-Solving Robot

Introduction

• Introduction of Jigsaw, a robot designed to solve jigsaw puzzles faster than humans.
• Potentially 200 times faster than the fastest competitive jigsaw puzzler.
• Discussion on the human abilities that aid in puzzle-solving.

Human Abilities in Puzzle Solving

1. Picking Up Pieces

• Human hands have 27 bones and 34 muscles for flexibility and precision.
• High concentration of nerves in fingertips for sensory feedback.
• Opposable thumbs enhance ability to manipulate objects.

2. Rotating the Piece

• Requires fine motor skills for correct orientation.

3. Moving the Piece

• Complex arm configurations allow for precise movement within a 3D space.
• Comparison of human arm configuration to other vertebrates.

4. Deciding Where the Piece Goes

• Subconscious processing of visual perception, pattern recognition, and spatial reasoning.
• Complex human brains enable quick decision-making.

Creating Jigsaw the Robot

Challenges in Replicating Human Abilities

• Task: Translate 200 million years of evolution into robotic capabilities.
• Breakdown of the four human tasks into robotic equivalents.

Steps to Robotize the Process

1. Picking Up a Piece
   • Use of a suction cup mechanism instead of opposable thumbs.
2. Rotating the Piece
   • Implementation of a precision donut motor for fine rotations.
3. Moving the Piece
   • Upgrading motors for high precision movement.
4. Deciding Where to Place the Piece
   • The most complex task; initially struggled with this aspect.

Breakthrough with Edge Analysis

• Collaboration with Ryan from Zipline led to a solution for edge analysis.
• Jigsaw uses a camera to take pictures and analyze edges for matches rather than relying on complex pattern recognition.

Edge Analysis Process

• Run a serpentine pattern over pieces for image collection.
• Convert edges into splines for comparison and matching.
• System to quantify mismatches and search for best fits among pieces.

Scaling to a 1000-Piece Puzzle

• Testing the robot’s capability on a larger scale puzzle.
• Initial success in solving a 12-piece puzzle before moving to the larger challenge.
• Time taken to analyze 1000 pieces was about one minute.

Final Assembly Challenges

• Identified sources of error in piece fitting due to slop and shifting.
• Jigsaw lacks sensory feedback to make small adjustments.

Solution to Fitting Errors

• Implementation of a z-height encoder to simulate human touch feedback.
• Development of a wiggle routine for precise fitting of pieces.

The Showdown: Robot vs Human

• Introduction of Tammy McLeod, a record-holding human puzzler.
• Initial competition with a simple 30-piece puzzle ended in human victory.
• A second competition with a 500-piece puzzle showed human strategies and techniques.

Key Takeaways from the Human Competitors

• Tips from Tammy for better puzzle-solving:
  1. Dump out and turn over all pieces.
  2. Assess edge piece strategy based on puzzle design.
  3. Grouping pieces by color, texture, or patterns.
  4. Sort similar shapes for easier matching.

Conclusion

• Final showdown between Jigsaw and Tammy resulted in Jigsaw triumphing in assembling a 1000-piece puzzle.
• Reflection on technological advancements vs. human capacity.
• Promotion of CrunchLabs Hack Pack for individuals interested in engineering skills and robotics.