Crosstalk-Suppressed Time-Stretch LiDAR Enabled by Random Pulsed Lasers for Massively Parallel 3D Detection

Time-stretch LiDAR, with its capability for multi-wavelength operation, is emerging as a critical pathway for high-speed 3D detection in multi-target scenarios, across both military and civilian domains. Nevertheless, a fundamental trade-off exists between the ranging resolution and channel number by increasing the number of phase-locked modes to compress the pulse width, coupled with the intense crosstalk due to disrupted frequency-to-time mapping during dispersive stretching when probing complex targets. These dual constraints pose significant challenges to system performance. Here, we proposed a crosstalk-suppressed time-stretch LiDAR that employed orthogonal spectra for wavelength-division multiplexing parallel detection, effectively overcoming the aforementioned limitations. The architecture integrated a highly nonlinear optical fiber into a partially mode-locked laser cavity, significantly enhancing spectral randomness to maintain high isolation between spectral channels. Based on Pearson correlation analysis for matching local and echo channels, we achieved a correlation peak width as narrow as 0.024 nm and demonstrated parallel detection of 250 channels. Directly extracting time-delay and intensity-difference information from the matched channel pairs allowed us to bypass the peak width limitation inherent in autocorrelation-based ranging, yielding spatial and depth resolutions of 4 mm and 3 mm, respectively, in parallel three-dimensional imaging. This work overcomes key constraints in large-scale parallel detection and provides a novel and robust solution for high-performance LiDAR systems.