Interference-robust waveform for LiDAR

Light detection and ranging (LiDAR) sensors provide accurate 3D vision to a vehicle, thus being fundamental for safe driving. However, currently available LiDARs rely on short pulse waveforms, which are permeable to interference. As more LiDAR sensors are deployed, interference between LiDARs shall be aggravated, jeopardizing reliable 3D vision and thus safe driving. In this paper, we propose a waveform that is robust against interfering pulses, while sacrificing as little range as possible. The proposed waveform consists of two pulses of known relative amplitude and position. A matching estimation method comprising a simple validation stage followed by a LightGBM-based estimation stage is presented. Such a method is demonstrated using an experimentally acquired dataset comprising interference-free signals as well as signals impaired by one interfering pulse. Over 97% of all interference-riddled signals are validated, and result in accurate estimations. In fact, estimations from interference-free signals are as accurate as estimations from a single pulse with the same average power, showing that resilience against interfering pulses was obtained without sacrificing estimation performance.