Rethinking Defect Detection in Modern Manufacturing with Matt Puchalski

From autonomous vehicles to factory floors, a new wave of vision technology is transforming how manufacturers think about quality. Bucket Robotics is at the center of that shift, bringing simulation-driven inspection systems to an industry long reliant on manual checks and outdated tooling.

Matt Puchalski, founder and CEO of Bucket Robotics, joins Greg to share how his experience in self-driving cars shaped a fundamentally different approach to quality inspection. Instead of relying on expensive hardware or months of data collection, his team is using CAD-based simulation to generate training data instantly, unlocking faster deployment, lower costs, and more scalable automation.

We explore why quality inspection remains one of the most painful bottlenecks in manufacturing, how legacy vision systems have failed to keep up, and what it takes to build robots that actually work outside of polished demos.

Highlights:

  • Matt’s journey from Georgia Tech and Michelin to autonomy startups and ultimately founding Bucket Robotics

  • Why quality inspection is still one of the most manual, inconsistent, and frustrating parts of manufacturing

  • The core insight behind Bucket: applying self-driving car vision systems to factory environments

  • How CAD-based simulation replaces months of data collection with minutes of synthetic training data

  • The “sim-to-real” challenge and why perception in changing lighting and environments is harder than it looks

  • Why most vision systems fail in production and how Bucket is designed for real-world robustness from day one

  • Lessons from early market assumptions, including why medical device manufacturing was not the right starting point

  • The economics of inspection: balancing cost, speed, and accuracy across high-mix and high-volume environments

  • What makes a strong customer fit, from ambiguous defect definitions to expensive rework caught too late

  • Common objections from manufacturers burned by legacy vision systems and how simulation changes the equation

  • Why labor shortages and supply chain reshoring are accelerating demand for automated quality solutions

  • Hiring for empathy in robotics and why understanding the end operator matters more than credentials

  • The importance of engineers who ship, not just prototype, and why early adopters beat bleeding edge thinkers

  • Hard-earned hiring lessons, especially the need for teams willing to travel and work onsite with customers

  • Where robotics is overhyped today, especially around deployment at scale versus polished demos

  • Why lightweight, lower-cost robotic systems are unlocking a new wave of practical automation

  • Matt’s view on the future of manufacturing: a hybrid human and robotic workforce rather than full autonomy

  • Founder reality: why building a company can feel easier than operating autonomous vehicles, but far more isolating

  • The long-term vision for Bucket Robotics as the “cloud computing moment” for manufacturing quality systems

Matt's LinkedIn: https://www.linkedin.com/in/matt-puchalski/

Bucket's LinkedIn: https://www.linkedin.com/company/bucketrobotics/

Matt's email: matt@bucketrobotics.com

Bucket's Youtube: https://www.youtube.com/@Bucket_Robotics

Connect with Greg Toroosian: https://www.linkedin.com/in/gregtoroosian/

Next
Next

Building Factory SuperIntelligence with Ariyan Kabir