iToF sensor survey

[omuryildirim]

To produce PointCloud, we need an accurate depth sensor with approximately 0.3m-4m range.

[omuryildirim]

ST's VL53L8

There are couple of eye catching sensors but to reduce processing power we started with a 8x8 resolution sensor from ST's VL53L8. The sensor has a 65 degree diagonal FoV.

Pros:

• Doesn't need any lens so pretty lightweight
• Doesn't require much power
• Easy to setup through I2C or SPI
• Easy to process data
• 60 Hz frame rate
• Proven that can be used for SLAM algorithms
• There are a lot of modules that can be purchased

Cons:

• Lower resolution with increased uncertainty with increased distance
• In order to run an SLAM algorithm like LIO-SAM, data is not rich enough
• Requires filtering and normalization on data

[omuryildirim]

ST's VD55H1

With it's high resolution and cheapness it is our main target sensor. Currently, there is no module in production that is using the sensor.

Pros:

• High resolution 672 x 804 pixel (0.5 MP)
• Can be used with low power high precision mode
• Up to 120 Hz frame rate
• Expected FoV is 108 degrees
• incredibly rich PointClouds
• Perfect for SLAM algorithms like LIO-SAM
• 0.3m-5m range

Cons:

• It's a camera sensor so it must be used with a lens
• A bit complicated setup due to MIPI CSI-2 quad lane or dual lane interface
• Needs at least 1.5GHz clock speed which makes it unusable with highly power efficient microchips
• There is no module in production yet

Potential module leads are [1, 2] which will be available in October.