Real-World Applications
& Case Studies
Exploring how our scalable Cartesian motion platform solves complex industrial challenges, from non-destructive battery testing to automated mass personalization.
Non-Contact Battery Ultrasound Scanning
The Challenge
Battery manufacturers face a fundamental inspection problem: the defects that matter most — internal delamination, voids, and micro-misalignments between cell layers — are completely invisible to standard visual inspection. These flaws sit beneath the surface, hidden inside the sealed cell structure, yet they’re often the earliest warning signs of thermal runaway, reduced capacity, or premature failure in the field.
Traditional contact-based ultrasound testing can find these defects, but introduces contamination risks from coupling gels and potential physical damage to the cells. Manual probing creates a severe manufacturing bottleneck with an error rate often exceeding 15%. The question: how do you achieve zero-defect battery production without ever touching the sensitive internal cell structure?
The Gantry Solution
We equipped a Cartesian gantry system with an air-coupled ultrasound transducer as the end effector — a fully non-contact, non-destructive testing method. No gels, no pressure, no physical interaction.
The gantry’s synchronized X-Y-Z motion control executes this at production speed. It maintains a constant, high-precision air gap between the sensor and the cell surface while running a boustrophedon (serpentine) raster pattern, guaranteeing continuous, systematic coverage.
Scanning Workflow
Initialization
Battery is loaded onto the fixture; the gantry moves to its starting corner.
Raster execution
Continuous X-axis loops with incremental Y-axis stepping cover the full cell surface.
Signal acquisition
Real-time pulse emission and reflected signal capture at every coordinate.
C-scan map generation
Acoustic data is reconstructed into a top-down defect map.
Defect flagging
Voids, delamination, and misalignments are automatically identified for rejection.
Why It Works & The Impact
Seamless Integration. Zero-Contact Inspection.
This system delivers sub-millimeter resolution defect maps at EV production speeds. Automated rejection catches failed cells instantly. What used to require slow, inconsistent manual probing now runs as a fully automated, 100%-coverage stage.
Non-destructive by design
Air-coupled transducers mean no gel and no applied pressure. Internal cell chemistry and mechanical integrity stay completely untouched.
100% surface coverage
The automated raster pattern guarantees every square millimeter is scanned — no sampling, no blind spots.
Production-line throughput
Consistent gantry motion speeds mean scanning keeps pace with high-volume EV battery production.
Industrial-grade repeatability
Cartesian path accuracy of ±0.05 mm and repeatability under 0.02 mm eliminates human variability entirely.
Full data traceability
Every coordinate links directly to its acoustic reading, generating a permanent digital twin for audit trails and compliance.
Technical Deep-Dive: The Physics and the Literature
Why air coupling is hard. The core physical obstacle is acoustic impedance mismatch — sound energy transmits poorly across the air-to-solid boundary. Published studies typically operate in the 350–450 kHz range. This is precisely why the gantry’s positional stability matters so much — any variation in the air gap directly degrades signal consistency.
What the scan is actually detecting. This method targets “stomata” or gas-pocket defects — internal gas generation that creates voids between layers. Measurable changes in the transmitted acoustic amplitude locate these defects regardless of depth.
C-scan as an industrial standard. Reconstructing a top-down defect map from an X-Y raster of point measurements is heavily used in aerospace composite inspection. Gantry systems resolving defects as small as 1 mm serve as a benchmark for this system.
Sealing and defect localization at the edge cases. Recent research uses leaky Lamb wave transmission at a critical incidence angle (around 22°) to inspect the notoriously difficult tab-to-pouch sealing interface, pointing toward future multi-angle scanning heads.
- Frontiers in Energy Research, Ultrasonic Tomography Study of Metal Defect Detection in Lithium-Ion Battery (2021)
- PMC/NCBI, Numerical Simulation and Experimental Study of Fluid-Solid Coupling-Based Air-Coupled Ultrasonic Detection of Stomata Defect of Lithium-Ion Battery (2019)
- PMC/NCBI, Air-Coupled Ultrasound Sealing Integrity Inspection Using Leaky Lamb Waves in a Simplified Model of a Lithium-Ion Pouch Battery (2022)
- Nondestructive Testing and Evaluation, Ultrasonic inspection of pouch-type lithium-ion batteries: a review (2024)
- QMI Inc. / TecScan, industrial C-scan gantry system technical documentation
Image-to-GCode Automated Laser Engraving
The Challenge
Laser engraving has traditionally been gated by a manual bottleneck: converting a design into a machine-executable toolpath requires CAD/CAM expertise. Every logo, photograph, or pattern has to be manually traced, pathed, and translated into GCode before a single line gets engraved — a process that can take hours per job and limits commercial viability.
The question driving this build: how do you eliminate the manual design bottleneck entirely and let the machine go straight from image to finished engraving?
The Gantry Solution
We built a vision-driven pipeline that takes a digital image as direct input and outputs precise machine motion — no manual pathing required. The gantry robot provides the high-precision X-Y planar travel needed for smooth, stable engraving, fitted with a diode/CO₂ laser module with dynamically adjustable power and focal control.
The Software Pipeline
Image input
Accepts photographs, logos, or vector artwork in JPEG, PNG, BMP, or SVG format — no pre-processing required.
Pre-processing
Automated contrast enhancement, edge detection, halftoning, and dithering translate grayscale into a physical format.
GCode generation
Pixel brightness values map directly to laser power (S-values), translating into standard G0/G1 linear movements.
Toolpath optimization
Raster passes are organized to minimize non-cutting travel time, tightly synchronizing motion and laser power.
Execution
The finished GCode streams to the gantry in real time for high-speed engraving.
Why It Works & The Impact
Mass Personalization at Scale.
Design-to-production time drops from hours of manual CAD/CAM work to minutes of automated processing — a 95% reduction in setup time. That shift turns one-off custom engraving from a slow specialist task into a scalable production capability across wood, acrylic, leather, and beyond.
Fully automated workflow
Manual GCode programming and CAD/CAM expertise are no longer required, reducing operator overhead.
High-fidelity reproduction
Advanced halftoning and dithering reproduce genuine photographic detail with fine tonal gradation.
Scalable customization
Software-driven scaling maps any image to any physical work area, making mass-personalized production viable.
Broad material compatibility
The system engraves cleanly across wood, acrylic, leather, anodized aluminum, stone, and rubber.
Technical Deep-Dive: The Image Science Behind the Pipeline
The core problem: too few tones, too much information. A laser engraver is effectively a bilevel output device. Digital halftoning generates a pattern of binary points from a continuous-tone image, exploiting the fact that the human eye doesn’t resolve high spatial frequencies.
Ordered dithering vs. error diffusion. Our pipeline defaults to error-diffusion dithering (like Floyd-Steinberg, Jarvis-Judice-Ninke, or Stucki algorithms) for photographic source material. These calculate rounding errors and propagate them to neighboring pixels, retaining much finer detail than ordered Bayer matrices.
Resolution trade-offs. Engraving resolution (DPI/LPI) is tuned per material and image by our optimization stage. Higher DPI captures detail but increases job time and char/burn risk.
Pixel brightness to laser power. Mapping pixel intensity to laser S-values (power) accounts for beam spot size, overlap, and maximum switching frequency. Proper calibration prevents muddy, low-contrast engravings.
- Jarvis, Judice & Ninke, A survey of techniques for the display of continuous tone pictures on bilevel displays (1976)
- Stucki, MECCA — a multiple-error correcting computation algorithm for bilevel image hardcopy reproduction (1981)
- Cloudray Laser, A Deep Dive into Dithering and Grayscale Processing (2025)
- Trotec Laser, Raster algorithms — helpful tips for laser engraving
- OMTech / LightBurn, Master Grayscale Engraving documentation
Application Comparison
A quick look at how the same LS-400 platform adapts to drastically different industrial requirements based entirely on end-effector and software changes.
| Feature Specification | Battery Ultrasound Scanning | Image-to-GCode Engraving |
|---|---|---|
| End Effector | Air-coupled ultrasound transducer | Diode/CO₂ laser engraving module |
| Motion Pattern | Serpentine raster scan | Image-derived raster/vector path |
| Automation Level | Fully automated execution | Fully automated (image → GCode) |
| Key Precision Metric | Scan position accuracy ±0.05 mm | Pixel-to-position spatial mapping |
| Primary Output | C-scan acoustic defect map | Engraved surface image |
| Material Interaction | Non-contact, non-destructive | Shallow surface material removal |