Structured light for robot vision: where it fits

Understand the measurement before buying the camera

Structured light is a way to capture surface geometry by projecting a known pattern and observing it with a camera. Zivid’s explanation of 3D vision principles describes the pattern-based approach and distinguishes it from other sensing techniques. For robot buyers, the useful output is geometry that can support the required decision: locating a part, choosing a grip, or checking a feature.

A dense point cloud is not itself a successful robotic application. The system still needs to identify the relevant object, decide whether a result is usable, transform it into the robot’s frame, and complete the intended task. Specify that chain before comparing attractive scan images from different suppliers.

Match the scene to the task

List the materials, finishes, colors, sizes, and presentation conditions the system will encounter. Include shiny surfaces, dark areas, recesses, edges, and parts that overlap. Record what the process actually needs to see. A hidden face cannot be measured merely because the camera has a high point count.

Ask the supplier to explain the proposed viewpoint and how the tool approaches the resulting target. Our pick-and-place planning page helps connect the scan with the physical handling. The sensor and gripper must be evaluated together: a visible surface may not be a usable contact, and a plausible grasp may still collide with the container.

Budget the entire acquisition and decision time

Different implementations use different capture sequences. Photoneo’s scanning guidance describes a static-scene requirement during projection for the scanners covered by that guidance. Confirm the motion limits and capture mode for the exact product under evaluation; do not apply one model’s rules to every structured-light system.

Time triggering, exposure or pattern sequence, data transfer, processing, robot communication, and any retry. If the part or camera moves, ask when that motion must pause and how the required position is confirmed. A camera-only capture figure does not establish the cell cycle. Include the worst expected part presentation in the timing trial.

Compare sensing options on the same samples

LMI Technologies’ measurement overview presents both structured-light and laser-triangulation approaches. Use such technical distinctions to choose candidates for testing, not to declare a universal winner. A stationary scene, moving profile, or long working distance may lead to different practical solutions once the whole cell is considered.

Our laser-profile scanner guide covers the alternative of building measurements along a moving part. Give competing suppliers the same sample set, required decision, spatial constraints, and timing limit. Ask for unedited results showing difficult examples and failures as well as successful scans. That evidence is more useful than demonstrations with unrelated objects.

Specify calibration and unusable-result handling

Request a coordinate-system diagram and a documented calibration process linking the sensor to the robot and workpiece. Identify what happens when the camera mount moves, the robot is serviced, or tooling changes. Keep calibration files and version information with the application backup.

If the system returns no valid target, it should follow an engineered response instead of reusing a previous pose without verification. Ask who controls retries, alternate views, operator intervention, and restart after a fault. Our camera integration guide separates software interfaces from these application responsibilities, while the TCP procedure governs the robot tool’s own reference.

Define acceptance evidence for the purchase

• Representative surface and presentation samples, including difficult conditions.
• Documented field of view, working distance, mounting, and environmental assumptions.
• Results tied to the required robot action or inspection decision.
• Measured full-cycle timing, retry behavior, and rejection handling.
• Calibration, maintenance, software, and support deliverables.

For a pilot, agree on how success and failure will be counted before collecting results. Separate a correct detection from a completed pick or acceptable inspection. Bring that trial plan to a vision-and-robot consultation. Purchasing structured light should be a decision about demonstrated application performance, not a contest over the most detailed-looking point cloud.

Scope structured light around a FAIRINO application

For a FAIRINO picking or inspection project, bring the representative sample set and required robot action into the same discussion. Select candidate arms from the FAIRINO lineup, then confirm camera placement, working distance, tooling, and the proposed controller interface for the complete cell. The FAIRINO selection guide helps connect those requirements to the quote. Request a FAIRINO vision-application quote with a trial plan that measures the completed task as well as the scan.