Abstract: A machine-learned (ML) model for detecting that depth data (e.g., lidar data, radar data) comprises a false positive attributable to particulate matter, such as dust, steam, smoke, rain, etc. The ML model may be trained based at least in part on simulated depth data generated by a fluid dynamics model and/or by collecting depth data during operation of a device (e.g., an autonomous vehicle. In some examples, an autonomous vehicle may identify depth data that may be associated with particulate matter based at least in part on an outlier region in a thermal image. For example, the outlier region may be associated with steam.

Abstract: A LIDAR system that identifies, from a channel output, a false positive return and/or suppressing a corresponding false positive detection caused, in some cases, a strong reflection by a highly reflective surface that caused light to leak from a first channel to a second channel. The LIDAR system described herein may identify, as a false return, a return detected in the second channel that has an intensity that is much less than a return in the first channel and indicates a distance that is the same or very close to a distance indicated the return in the first channel. Based at least in part on identifying a return as a false return, the LIDAR system may suppress a false detection associated with the false return by modifying a detection threshold.

Abstract: Particulate matter, such as dust, steam, smoke, rain, etc. may cause one or more sensor types to generate false positive detections. In particular, various depth measurements may be impeded by particulate matter. Identifying a false return and/or removing a false detection based at least in part on a sensor output may comprise determining a similarity of a portion of a return signal to an emitted light pulse or an expected return signal, determining a variance of the signal portion over time, determining a difference between a power spectrum of the return relative to an expected power spectrum, and/or determining that a duration associated with the signal portion meets or exceeds a threshold duration.

Abstract: Particulate matter, such as dust, steam, smoke, rain, etc. may cause one or more sensor types to generate false positive detections. In particular, various depth measurements may be impeded by particulate matter. Identifying a false return and/or removing a false detection based at least in part on a sensor output may comprise determining a similarity of a portion of a return signal to an emitted light pulse or an expected return signal, determining a variance of the signal portion over time, determining a difference between a power spectrum of the return relative to an expected power spectrum, and/or determining that a duration associated with the signal portion meets or exceeds a threshold duration.

Abstract: A machine-learned (ML) model for detecting that depth data (e.g., lidar data, radar data) comprises a false positive attributable to particulate matter, such as dust, steam, smoke, rain, etc. The ML model may be trained based at least in part on simulated depth data generated by a fluid dynamics model and/or by collecting depth data during operation of a device (e.g., an autonomous vehicle. In some examples, an autonomous vehicle may identify depth data that may be associated with particulate matter based at least in part on an outlier region in a thermal image. For example, the outlier region may be associated with steam.

Abstract: A LIDAR system that identifies, from a channel output, a false positive return and/or suppressing a corresponding false positive detection caused, in some cases, a strong reflection by a highly reflective surface that caused light to leak from a first channel to a second channel. The LIDAR system described herein may identify, as a false return, a return detected in the second channel that has an intensity that is much less than a return in the first channel and indicates a distance that is the same or very close to a distance indicated the return in the first channel. Based at least in part on identifying a return as a false return, the LIDAR system may suppress a false detection associated with the false return by modifying a detection threshold.