AMR Navigation in Dust, Smoke, Glass, Sunlight, and Welding Arc: Failure Modes and Acceptance Testing

The navigation accuracy on an AMR spec sheet is measured under ideal conditions: flat floor, adequate lighting, no dust, no smoke, no reflective surfaces, no dynamic obstacles.

Real factory or warehouse optical environments are rarely ideal. Dust, smoke, glass walls, direct sunlight, welding arc — any one of these can cause navigation sensors to misjudge or fail. Buyers have reported “navigation unreliable, no positioning sustained” messages when floor texture degrades — this is not a device fault but a real manifestation of environmental conditions affecting visual navigation feature loss.

This article focuses on navigation failure modes in adverse optical environments, symptom identification, and acceptance testing methods. It does not cover SLAM basics.

Dust and Smoke: False Obstacles and Feature Loss

Dust impact on navigation sensors:

Smoke impact: High concentrations of airborne particles or smoke can degrade optical sensing performance; the degree of degradation depends on sensor type, wavelength, filtering, installation, and concentration, so the actual system must be tested under representative conditions.

Mitigation strategies:

• Periodic sensor cleaning — frequency depends on dust concentration.
• Path planning to avoid high-dust areas — if possible.
• Multi-sensor fusion — degradation behavior is system-specific; verify what happens when input quality degrades.
• Speed reduction — lower speed increases reaction time.

If your environment has dust or smoke (near cutting, grinding, welding operations), ask the supplier to run navigation tests under these conditions.

Glass, Mirrors, and Reflective Metal

Reflective surface impact on navigation sensors:

Buyers have reported laser navigation failures near glass walls — false detection points prevent the robot from passing normally.

Mitigation strategies:

• Path planning to avoid glass and mirrors — if possible.
• Apply opaque markers on glass — provide trackable features for visual sensors.
• Multi-sensor fusion — degradation behavior is system-specific; verify with the supplier what happens when one sensing modality degrades.

If your route has glass doors, mirrors, or large reflective metal surfaces (stainless steel equipment), ask the supplier to run navigation tests under these conditions.

Direct Sunlight, Shadows, and Welding Arc

Strong light impact on navigation sensors:

Shadow impact: Shadow boundaries cause visual feature discontinuities — the robot may lose positioning at the sun/shade boundary.

Welding arc: The high-intensity arc in welding areas contains strong UV and visible light — within close range of welding stations, the intense arc may degrade optical sensor performance. If the AMR route passes near welding stations, verify the sensing architecture and site conditions with the supplier.

Mitigation strategies:

• Direct sunlight areas: Select a sensing architecture qualified for the environment — verify with the supplier what sensing options are available for high-ambient-light conditions.
• Shadow boundaries: Mark these zones in path planning, reduce speed through them.
• Welding areas: Verify the sensing architecture and site conditions with the supplier; any route restrictions must follow the site safety and process plan.
• Outdoor or semi-outdoor routes: Consider sun direction and time-of-day effects.

If your route has direct sunlight (near windows or outdoor sections) or welding operations, ask the supplier to run navigation tests under these conditions.

Repetitive Geometry and Low-Feature Areas

Low-feature areas: Long corridors, large white walls, repetitive shelf rows — these areas lack trackable features, and navigation sensors may “get lost.”

Impact:

• LiDAR SLAM: All scan points look similar → cannot distinguish position → positioning drift.
• Visual SLAM: Image features repeat → feature matching errors → jump to wrong position.
• QR code / marker navigation: Does not rely on environmental features → unaffected.

Mitigation strategies:

• Add reference markers in low-feature areas (reflectors, QR codes, feature patterns).
• Multi-sensor fusion — IMU can maintain positioning for short periods.
• Mark low-feature zones in path planning, reduce speed through them.

If your warehouse has long corridors or repetitive shelf rows, confirm the navigation solution’s positioning reliability in low-feature areas.

Safety Sensing vs. Navigation Sensing: Boundary Note

A critical distinction: safety sensors (safety LiDAR, safety light curtains, e-stops) and navigation sensors (2D/3D LiDAR, cameras, IMU) are separate systems with different purposes. Adverse optical environments affect both, but the consequences differ — navigation sensor failure causes operational impact (lost positioning), while safety sensor failure causes safety impact (false stop or missed detection).

Any protective-field or safety function adjustment must follow the manufacturer’s procedure and the site safety validation. This article does not provide safety sensor design or configuration advice — safety function selection, verification, and adjustment are covered in the dedicated safety sensing article.

Sensor Fusion and Graceful Degradation

The above are single-sensor failure modes, but real systems use multiple sensors in coordination — understanding fusion and degradation strategies is key to evaluating navigation robustness.

Multi-sensor fusion is not “more sensors = safer” — fusion behavior is system-specific. Verify what happens when input quality degrades: does the system reduce speed, switch to a different sensing mode, or stop? The specific degradation logic depends on the system’s software architecture and configuration.

Fusion strategies:

Graceful degradation: The system may reduce performance when a sensor’s input quality degrades — the actual behavior depends on the system’s fusion and degradation logic. This requires degradation logic designed into the software.

Confirm each sensor’s role in the fusion scheme — who is primary, who is backup? How does it degrade on failure? What is the post-degradation performance expectation?

What to Record: Lost Positioning, False Stops, and Manual Recovery

The operational impact of navigation failure is not just “how many mm did accuracy drop” but also these metrics:

Environmental Navigation Acceptance Test Matrix

Environmental Test Inputs

1. Spec sheet accuracy is under ideal conditions — your environment will almost never fully match.
2. Dust, smoke, glass, strong light, and low features are the five navigation killers — each has different failure modes.
3. Safety sensing and navigation sensing are separate systems — safety field adjustments are not a navigation tuning exercise and must be done by qualified personnel.
4. Sensor fusion behavior is system-specific — verify what happens when input quality degrades, do not assume automatic takeover.
5. Safety sensing is a separate concern — any protective-field or safety function adjustment must follow the manufacturer’s procedure and site safety validation.
6. Low-feature areas need additional markers — long corridors and repetitive shelves are SLAM blind spots.
7. Acceptance is not just about mm — it is about false stops, lost positioning, and manual recovery frequency — these operational metrics reflect real usability.
8. Test in adverse optical environments at both FAT and SAT — clean-floor demos cannot replace real-environment verification.