Feasibility of VL53L5CX / VL53L8CX for Outdoor Motion Tracking Application

[TRAXTech]

Hello everyone,

I’m evaluating ST ToF sensors (VL53L5CX / VL53L8CX) for an outdoor motion-tracking use case and would like to understand the general feasibility before investing in prototypes.

Setup:
A ToF sensor would be placed approximately at ankle height behind a moving person, facing upwards. The goal is to detect center-of-mass shifts and general arm/torso positioning while the person moves — not detailed skeletal joints, just coarse body-position indicators.

Environment:

• Potentially strong sunlight and ambient IR exposure

• Possible water droplets, spray, and wet surfaces near the sensor (but not submerged)

• Target surfaces are typically dark fabrics or wetsuit-like materials at a distance of 0.5–2 m

 

Questions

1. General feasibility – Can VL53L5CX or VL53L8CX reliably detect human body shape or coarse motion under these lighting and environmental conditions, or would performance degrade too much in direct sunlight and with occasional droplets on the cover glass?

2. Environmental tolerance – Are there known limitations or mitigations for operation near moisture (e.g., hydrophobic coatings, tilted windows, or protective housings) that can keep the sensor functional outdoors?

3. Resolution needs – The built-in 8×8 zone grid is appealing, but would that be sufficient to distinguish major limb/torso motion at ~1–2 m? Or should we expect that only higher-resolution ToF (e.g., TMF8829-class or camera-grade depth sensors) can meaningfully separate these shapes?

4. Recommended starting point – For evaluating outdoor motion-tracking feasibility, would you recommend beginning with VL53L5CX or VL53L8CX, and are there evaluation boards better suited for testing under bright IR and wet conditions?

I’d really appreciate the team’s perspective on whether multi-zone ToF sensors can realistically capture large-body movement outdoors, or if the physics and environment make this fundamentally unreliable.

Thank you!

Michael

[John_Kvam]

First, the easy one. Only consider the VL53L8. It is in all respects a better sensor than the VL53L5 and the cost is the same. The L8 has better lenses created by printing them the same way ST prints the silicon wafers. So, the lens is cheaper, and the laser power is greater. 

The question is - will it help it overcome very bright sunlight? And to be honest I do not know. 

What I do know is the worst case is the sun directly behind the sensor. As the sensor is looking up from the ground, that's not such a big deal. And if you know the body is between the sensor and the sun, you might get away with it - even on a bright sunny day. 

But at 2 meters you might only see the body mass and not hands or arms. 

When water gets on the coverglass, you are going to increase crosstalk. But on the L8, crosstalk only affects the first 60cm or so. After that the sensor can tell the difference between crosstalk and the actual target. So if the body is more than 60cm away, it should not be that much of an issue. Under-estimating the crosstalk will cause the range to appear shorter than actual. But if you know the minimum expected distance you might be able to work out the correct distance. 

Clothing is tricky. If you told me that all your targets wear wool, I'd tell you not to worry as wool - no matter what color it's dyed - is very reflective. Cotton fleece, on the other hand - is not very reflective. Again, dye color does not matter. 

Plastics - Rayon, Nylon, polyester etc - are all somewhere in the 50% reflective range.  If you want to guarantee success, have your test subjects wear safety vests. That retro-reflective tape is VERY reflective. 

 - john 

[TRAXTech]

John, thanks—that’s super helpful. Two quick follow-ups:

1. Target reflectivity (neoprene / bare skin):
   In your experience, how do dark neoprene wetsuits compare to bare skin for the VL53L8 in outdoor use? Any noticeable difference when the surface is wet vs. just damp? If you’ve seen specific pitfalls with neoprene (e.g., very low return, glancing angles), any mitigation tips?

2. Detecting arms (range & geometry):
   You mentioned that at ~2 m we may only see the body mass. Is there a practical range window (e.g., 0.8–1.5 m) or viewing geometry/FOV where arms become distinguishable from the torso with the L8’s 8×8 grid? If yes, would you recommend narrowing the ROI, changing the mounting angle, or adjusting frame rate/VCSEL power to improve separation?

Really appreciate your guidance!