VSLAM relies on extracting and matching visual features between overlapping camera images. Certain environmental conditions reduce the number or quality of features and can cause poor behaviour or loss of tracking.
Conditions that can cause VSLAM to malfunction:
- Absence of overlapping images - moving too fast or leaving gaps between passes provides too little overlap for feature matching.
- Absence of features in the object - low-texture or featureless surfaces give the algorithm too few points to track.
- Dark objects or insufficient lighting - too little light reduces feature extraction and degrades VSLAM behaviour.
- Over-saturated lighting or glare - shiny objects and over-exposure wash out features and impact camera visibility.
- Poor visibility or turbidity of the water - suspended particles obscure the scene and reduce usable features.
- Excessive speed - movement faster than 0.2 m/s causes motion blur and insufficient overlap, leading to track loss.
How to reduce these issues: keep a steady distance of 0.5 to 1 m from the structure, move at no more than 0.2 m/s with a stable attitude, ensure even and sufficient (non-saturated) lighting, plan good overlap between passes, and avoid highly turbid water where possible. If tracking is lost, hold position and return to the last tracked area to relocalise.
Note: Full operating recommendations are described in the Voyis VSLAM Powered by EIVA NaviSuite Quick start guide.