AI Guided Welding Goes Mobile with Spot

A welding torch now roams the shipyard on four legs, guided by AI.

Path Robotics is applying physical AI to a problem that has long defied automation: welding. Testing shows the company’s system can identify the torch path and move the robot through a seam with real time vision, keeping the arc on target even as joint lines shift or fixtures are reconfigured. The approach treats welding as a sensing and control problem rather than a fixed set of motions, a shift that could reshape how shops think about programming and maintaining weld lines in production.

Andy Lonsberry, co-founder and CEO of Path Robotics, describes the effort as building adaptive, AI driven welding cells designed for real world production environments. The Columbus, Ohio based startup has framed its stack around learning representations that connect perception to motion, letting robots respond to changes in joint geometry rather than requiring new offline programs for every part. Testing shows the AI can consistently identify the torch’s path and steer the robot to follow it, a capability that cuts the guesswork out of setup and reduces reliance on highly skilled manual programming.

In a broader deployment push, Path Robotics is also bringing mobile welding capabilities to shipyards by deploying Boston Dynamics Spot quadruped robots as roaming welders. The company reports that Spot is being used to carry welding heads to hard to reach areas and to reconfigure work cells on the fly, expanding where automation can operate within large, cluttered environments like ships under construction. This combination of vision guided welding and mobile platforms signals a move beyond static welding cells toward flexible, reconfigurable manufacturing lines.

From a practitioner’s perspective, the AI driven path optimization promises several practical gains. Testing shows real time guidance can adapt to variations in joint geometry and fixture placement, potentially reducing rework from misaligned welds. The integration with Spot introduces new capabilities for access and coverage, enabling welds to be performed in awkward or constrained locations where fixed robots could not reach. The approach also raises important considerations: perception fidelity in harsh shipyard conditions (smoke, glare, and dust), the need for robust calibration between vision systems and robotic arms, and the importance of reliable safety protocols when a mobile robotic welder shares space with human workers.

The company’s framing here is explicit about the benefits and the constraints. Documentation indicates that AI based welding relies on real time feedback to adjust the torch path rather than preprogrammed coordinates alone, a design choice that helps cope with deviations common in production lines. The Robot Report Podcast coverage underscores that the challenge remains translating lab grade perception into noisy, dynamic factory floors, but Path Robotics is betting on this feedback loop to improve weld consistency and throughput in practical settings.

If the approach scales, expect a tighter relationship between sensing, control, and mobility in welding workflows. Two to four concrete practitioners’ watch points emerge: ensure robust perception under industrial lighting and smoke, maintain reliable synchronization between the vision stack and the robot controller, plan for maintenance and calibration cycles that keep the AI stack aligned with target welds, and build safety layers that govern a mobile welder’s interactions with human teammates and other machinery in the yard.