Modernizing Robotic Laser Peening with Simulation

At the Curtiss-Wright Surface Technologies facility in California, a team of engineers faced a persistent challenge. Their laser peening operation, which improves fatigue life and crack growth properties on metallic parts, relies on multiple robots working in precise choreography: One robot holds the part, another directs a laser beam via mirrors, and one robot positions water jets to confine the plasma during laser treatment. Each robot must reach its designated position precisely without interfering with the others.

Curtiss-Wright engineers relied on a manual, iterative process to program their robots, according to Vincent Sherman, a process engineer at Curtiss-Wright Surface Technologies.

Even after the initial programming is complete, additional problems can occur during operation. For example, the water jets gradually shift position, based on a variety of environmental factors.

“A very small shift in orientation, from the tool getting bumped or a small error in scanning, could create a large movement on the robot wrist that makes the intended position unreachable if we’re not careful,” says Sherman.

When issues arose, engineers would need to step in, recalibrate the tool, and adjust robot targets to ensure everything worked—a process that consumed valuable production time and engineering resources. These disruptions would occur several times per week, resulting in significant downtime.

Digital Simulations Prevent Work Disruptions

This approach has not only made the team’s daily operations more reliable but has also significantly reduced troubleshooting time during production. Validating positions before starting work has virtually eliminated the mid-process stoppages that previously disrupted their flow.

Complex Collision Modeling

One key challenge in creating accurate simulations was developing detailed collision models for the robots and aircraft parts, many of which have complex geometries with concave surfaces.

“When our robots go inside an aircraft, a careful evaluation of their operation is safety critical,” says Sherman.

The team balances computational efficiency with safety. Team members use simpler models in open areas. For complex areas like aircraft interiors, they leverage convex decomposition capabilities in Robotics System Toolbox that represent the actual shape of components rather than simplified approximations, ensuring advanced collision detection in tight spaces.

For areas with simple geometries, the system uses convex hull representations—essentially simplified envelope models of the structures. But in complex spaces where precision matters most, more detailed, advanced decomposition techniques that accurately model every contour and surface are used. This selective approach to collision modeling means the system focuses the computing power where it’s most needed.

The system’s ability to import and accurately model both the team’s robot configurations and the complex workpieces they’re processing has been crucial to the success of the simulation. The system can detect potential collisions that a simpler modeling approach might miss, particularly in the confined spaces where Curtiss-Wright’s robots operate.

This prevents not only damage to expensive equipment but also potential safety hazards and lengthy production delays that would occur due to a collision during operations.

The Ripple Effect of Simulation

This acceleration is particularly noticeable with complex parts such as crankshafts for diesel engines. The rounded shape of these components requires precise positioning of the laser tool to achieve proper peening angles, a task engineers found particularly challenging before simulation. The inverse kinematics (IK) solver enabled a direct and rapid solution, providing precise and accurate robot positioning.

Beyond path planning, the team now uses simulation to validate laser beam paths through ray tracing. This allows them to resolve the exact path a laser will take from the tool to the part, ensuring there are no collisions or obstructions that would interfere with the process. It also helps prevent damage to the costly mirrors used to direct laser light onto the workpiece.

“If too much of the laser energy is passing through too small an area, then in one shot, we’re going to destroy the mirror,” says Sherman. “This ray tracing helps us protect them.”

The simulation tools have also improved communication with customers. When they question why Curtiss-Wright needs clearance around parts, Sherman can now demonstrate exactly how the laser beam will travel. Disassembling aircraft for access is costly, so specific and intentional preparation saves time and money. Previously, customers would question the cost-effectiveness of clearing space around the target area, while the ray tracing makes the need for preparation clear and obvious.

“We get to show them, ‘Well, this is why. This is the path the beam is going to take,’” says Sherman.

Expanding Applications in Data Analysis and Quality Control

Data analysis has also become a critical component of Curtiss-Wright’s MATLAB implementation. The system collects extensive data during the laser peening process, including energy readings, pulse widths, and other measurements for every laser shot. A wealth of information accumulates quickly, with daily logs reaching up to one million lines of text, an amount engineers struggled to utilize effectively. Now, custom MATLAB applications allow engineers to analyze and visualize these massive data sets.

“The ability to analyze, visualize, and work with data has let us understand in the long term what the strengths and weaknesses of our systems are,” says Sherman.

A Growing Toolset for Future Innovation

For the engineers at Curtiss-Wright, the simulation-based approach represents a fundamental shift in how they work with robotic systems—one that will continue to evolve as customer demands grow more complex.

“What’s driven us to get here is our customers asking us for more: greater volume, highly complex parts, and rapid process development,” says Sherman. “This is indefinitely going to drive the level of the analysis and simulations that we need.”