User Guide

54A0049_7.0_EN / Leddar Pixell – User Guide 19 | 127
2.1. Optical Crosstalk
Lens flare, diffusion, and reflections may cause undesired signals, which are commonly referred to as optical
crosstalk. To resolve this issue, LeddarSP uses a method that is based on a deconvolution process to
reverse these effects.
The platform approach has been adopted in LeddarSP, and two chosen methods have been carefully
implemented.
The first method is based on pulse deconvolution; it has the advantage of requiring a very short computation
time, which optimizes the addressable pixel per second. The second method is based on trace deconvolution
and requires more computing time.
Both methods’ performance improvement on false detections is limited by calibration imperfections.
2.2. Electronic Crosstalk
Additionally, LeddarSP corrects photodetector and electronic imperfections in the detection. When a light
pulse is received on a photodetector channel, artefacts are created in the electric signal that is generated by
the adjacent photodetectors, which is commonly known as electronic crosstalk (“xtalk”).
The shape and amplitude of the electronic crosstalk depend mainly on the sensor’s characteristics.
The amplitude of an electronic crosstalk signal is about 40 dB below the aggressor amplitude for the pixels
just next to the aggressor, decreasing gradually to reach about 60 dB from 4 pixels from the aggressor.
Both optical and electronic crosstalks are mitigated by calibration measurements in specially designed
manufacturing set-ups.
The Leddar Pixell embeds LeddarEngine algorithms designed to mitigate electronic crosstalk and minimize
the probability of spurious detections, providing an improvement of more than 16 dB on amplitude
measurements.
The exact performance of LeddarSP algorithms varies from unit to unit and with the scene view by the LiDAR.
The reflectivity, size, and distance of objects to the LiDAR and object proximity to other objects impact the
ability of the global LeddarSP algorithms to mitigate electronic crosstalk.
Therefore, with extremely strong aggressors, it is still possible to have a significant residual after a correction.
This may lead to unwanted detections (i.e., false positives) or missed detections (i.e., false negatives).
However, the detections in an area that have been corrected are flagged (see Table 27 on page 79); they
must, therefore, be used with more caution.
The corrected area, where false positives and false negatives can occur, extends over 17 meters around the
aggressor at most. The start and the end of a correction depends on the context; the start can be up to
7.7 meters before the aggressor, whereas the end can be up to 14.4 meters after the aggressor.
2.3. False Positives and False Negatives Around Reflective Events
Both modes of crosstalk inject pulse-like signals on the same photodiode array and contaminate a good
number of lateral segments. Special algorithms are used to remove false positives on the same line and both
sides around the strong reflector. Depending on the intensity of the reflective event, the capacity of detection
on the horizontal line is limited up to 7 meters behind the reflector.