Frequency Modulated Continuous WAVE Linearization of LiDAR Using Reinforcement Learning

The frequency-swept laser (FSL) acts an important part in frequency modulated continuous wave (FMCW) light detection and ranging (LiDAR), while its inherent nonlinearity is always deemed as one of the most important issues. To address this weakness, we import deep reinforcement learning (RL) to control the modulation signal of the FSL to guarantee the linear FMCW generation. The experimental result shows that the distance resolution raises more than one order of magnitude with the RL control, indicating that our novel routine is excellent for the feedback control to realize the purpose of the FMCW linearization.