The advanced equipment manufacturing industry, represented by industrial measurement equipment calibration, high-precision 3D control network construction, and large-scale assembly pose detection, has put forward demands for high-precision, versatile, and multi-dimensional geometric measurement in measurement technology^[1–5]. Multi-laser ranging technology demonstrates outstanding performance, such as unified sensing mode, minimal error sources, and high measurement efficiency. Moreover, it also holds significant advantages in enhancing the measurement accuracy of individual targets and achieving comprehensive multi-dimensional measurements of targets^[6,7]. Frequency modulated continuous wave (FMCW) laser measurement technology, due to its technical advantages, such as high precision, large dynamic range, strong anti-interference capability, and high signal-to-noise ratio, can achieve precise measurement of absolute distances for large-scale, non-cooperative targets^[8–11]. This coherent measurement method exhibits greater robustness to environmental variables and phase noise compared to traditional LiDAR technologies represented by time-of-flight (TOF) technology^[9,12,13]. Therefore, FMCW LiDAR demonstrates enormous potential in industrial measurement applications, such as precision machining, equipment assembly positioning, and multi-dimensional shape scanning.

The characteristics of traditional LiDAR systems include large volume and high cost. The continuous maturation of integrated photonics technology has led researchers to foresee the dawn of integrating and chylifying LiDAR^[14,15]. The microwave photonic radar with chip-scale design achieves high-precision absolute distance measurement with a resolution of 2.7 cm and an error of less than 2.75 mm^[16]. Micro resonators integrated on a silicon-based chip enable calibration of laser modulation nonlinearity, reducing the volume of FMCW LiDAR while enhancing measurement accuracy^[17]. Utilizing a monolithic integrated linear frequency-modulated dual-wavelength distributed feedback laser (DFB) enables high-frequency linear scanning and demonstrates precise measurement of single-mode optical fibers, improving measurement efficiency and repeatability^[18,19]. It is worth noting that the chip-scale LiDAR solutions mentioned above commonly employ mechanical scanners such as oscillating mirrors and microelectromechanical system (MEMS) scanners to expand the imaging range^[20,21]. However, this greatly limits the efficiency of the FMCW LiDAR with large time-bandwidth product (TBWP).

A large-scale parallel coherent measurement architecture can significantly improve measurement efficiency by enabling multichannel high-rate data acquisition without relying on mechanical beam scanning devices. In 2020, a team from the Swiss Federal Institute of Technology proposed a large-scale parallel coherent LiDAR scheme based on micro resonator soliton optical frequency combs (OFCs). Due to the relatively large repetition frequency of the micro-resonator soliton optical frequency combs, each comb tooth can be used for frequency-sweep interferometry (FSI) ranging when the optical frequency of the pump source is linearly modulated. This enables the parallel application of FMCW LiDAR, with a ranging accuracy of less than 1 cm for 30 detection channels^[2]. In 2023, a team from Peking University implemented a 31-channel FMCW LiDAR using cascaded electro-optic modulators to generate a frequency comb. Each channel of this LiDAR system achieves a high scanning bandwidth of up to 15 GHz, corresponding to a measurement resolution of 1 cm, with a measurement rate exceeding 12 million pixels per second^[22]. However, these efforts require complex laser designs and system architectures. The number of measurement channels determines the quantity of components, such as modulators, photodetectors, and optical amplifiers, significantly increasing the complexity of the system. Furthermore, existing parallel coherent measurement architectures still suffer from relatively small modulation bandwidths for each channel in the FMCW LiDAR. Non-mechanical beam scanning devices offer limited imaging range with fixed paths and are unable to meet the demands of industrial applications for high-precision and multifunctional measurements on large-scale and non-cooperative targets.

Here, we propose a novel multichannel time-domain wavelength division multiplexing (TWDM) FMCW LiDAR system integrated with the optical switch for versatile distance measurement and 3D imaging. This system combines the large TBWP modulation characteristics of the FMCW LiDAR with the continuous temporal band switching capability of the optical switch to achieve time-domain wavelength division multiplexing technology. We employ the K-domain triggering resampling method to mitigate laser modulation nonlinearity and nonlinear errors introduced by oscilloscope sampling, optimizing the accuracy of absolute distance measurements^[8,23,24]. To reduce system errors, we utilize a Fizeau common-path interferometer structure along the measurement path. Different from the multi-line technology, this system features fast measurement speed, high resolution, good stability, and a simple structure. It can be applied to multifunctional high-precision measurements and 3D imaging. In typical industrial precision measurement scenarios, such as machine tool displacement measurement and pipeline deformation monitoring, space constraints and complex environmental disturbances often limit the working distance to a range of 0.5–2 m. Against this backdrop, we demonstrated the application of the TWDM-FMCW LiDAR system for non-cooperative single-target multichannel measurements and overall measurement accuracy improvement. For targets at distances ranging from 1.2 to 1.36 m, the maximum overall measurement error did not exceed 14 µm, and the measurement accuracy of each channel remained within a range of 20 µm. The overall standard deviation reached 14.73 µm, and the minimum Allan deviation was 189 nm at a 2.84 s averaging time. Furthermore, by combining with an auxiliary scanning system, we conducted 3D imaging experiments on paper with hollow patterns and stairs, demonstrating the effective acquisition of high-precision, actual 3D imaging results. The TWDM-FMCW LiDAR system exhibits multifunctional and high-precision measurement capabilities, providing a benchmark for large-scale spatial measurements and multidimensional geometric measurements in industrial scenarios.

FMCW LiDAR uses frequency chirping to map distance to frequency. By combining the beat frequency information obtained from the interference between the echo signal and the local oscillator signal, the target distance can be calculated. The modulation methods of the FMCW lasers are generally set to triangular and sawtooth waves. We propose an FMCW LiDAR system architecture that integrated with the optical switch. This architecture leverages the large TBWP continuous modulation characteristics of FMCW lasers and optical switch technology to achieve multichannel continuous time-domain wavelength division multiplexing, enabling multifunctional high-precision ranging and 3D imaging. The schematic diagram of the multichannel TWDM-FMCW LiDAR system architecture and the measurement process is shown in Fig. 1(a).

The FMCW laser is divided into two paths, simultaneously entering the measurement optical path and the local oscillation optical path. The laser in the measurement optical path is amplified by the EDFA, then enters a circulator, and subsequently is directed into a controllable optical switch array. The optical switch array we use is based on MEMS technology, employing optical micromirrors to manipulate the direction of the light beam and switch optical paths. Small mirrors are etched onto a silicon substrate, and the movement of the mirrors is achieved by applying voltage or magnetic fields to electrostatic actuators integrated into the mirror structures. The principle of the MEMS optical switch and its actual photos are shown in Fig. 1(d). Under the continuous time-domain wavelength switching effect of the optical switch array (with a switching time of 300 ns per channel), the laser is divided into multiple non-overlapping measurement channels. Each wavelength of the laser is directed towards the free-space target after passing through a fiber collimator. The local oscillation optical path laser is split into two paths by a 50/50 beam splitter: one path constructs an auxiliary path through a delay fiber, and the other serves as a reference light that interferes with the returned light from the measurement path and the emitted light from the auxiliary path. The interference signals are received by a balanced detector and collected using an oscilloscope. Figure 1(b), respectively, shows the measurement signals collected by the digital oscilloscope before and after time-domain wavelength division multiplexing. Compared to the traditional FMCW LiDAR system, which resolves a single distance information per cycle, the multichannel TWDM-FMCW LiDAR system can resolve continuous multi-distance information per cycle.

The principle of FMCW laser measurement is shown in Fig. 1(c). Taking the single cycle frequency sweep process of the triangular wave frequency modulation process as an example, the instantaneous electric field corresponding to the local oscillator light is Em(t)=Aexp(i(2π(f0t+αt2/2)+φ0)),0<t<T/2,(1)where f0 is the initial frequency of the laser modulation, α is the chirp rate of the laser wavelength, T is the modulation period of the laser, and φ0 is the initial phase. The amplitude is denoted as a constant A.

The local oscillator light is emitted from the collimator to the target, reflected to the collimator, and the returned light has a time delay τ relative to the local oscillator light. The instantaneous electric field corresponding to the returned light currently is Er(t)=Aexp(i(2π(f0(t−τ)+α(t−τ)2/2)+φ0)),0<t<T/2.(2)

The local oscillator light and the returned light interfere, random phase is ignored, and the instantaneous electric field intensity of the interference beat signal is $$\begin{gathered} Im(t)=(Em(t)+Er(t))×(Em(t)*+Er(t)*) \\ =Im+Ir+2ImIrcos(2π(ατt+f0τ+ατ2/2)),0<t<T/2, \end{gathered}$$(3)where Er(t)* and Em(t)* are the conjugate forms of the optical fields of the local oscillator signal and the measurement signal, respectively. Ir and Im are the optical intensities of the local oscillator signal and the measurement signal, respectively.

After filtering out the DC component, the signal becomes a standard cosine signal with frequency represented as fm=ατ. The target distance can be expressed as R=cτm2n=cfm2nα,(4)where τm represents the measurement distance time delay, c represents the speed of light, and n represents the refractive index of the air.

We utilize the optical switch to divide the wide-band continuous modulated light with a large time-bandwidth product into multiple non-overlapping measurement channels, achieving the time-domain wavelength division multiplexing technology for FMCW LiDAR, as shown in Fig. 2. The multichannel interference signal at this time can be expressed as I(δ)=∑i=1nIr0(δi)×cos(2π(ατiδi+f0τi)),(5)where i ranges from 1 to n, representing each detection channel, δi represents the time length of each channel in the time-domain wavelength division multiplexing, Ir0(δi) represents the intensity of the interference signal corresponding to each channel, and τi represents the ranging delay generated by each channel.

Under a unified trigger clock, this method enables nearly synchronous measurements of multiple measurement positions (targets), and it also allows for improving the accuracy of single measurements by averaging multiple measurements for the same target, thereby demonstrating the multifunctionality of the TWDM-FMCW LiDAR. In this case, the distance measurement by the FMCW LiDAR for the same target is R=c2nM∑i=1Mτi=c2nM∑i=1Mfiα,(6)where i represents the ith group of signals in time-domain wavelength division multiplexing, M represents the number of channels, α is the chirp rate of the laser wavelength, and fi represents the beat frequency for the ith group of signals.

In practice, frequency modulation of FMCW lasers often exhibits nonlinearity, with the laser output containing a nonlinear modulation term fe(t). In this case, the signal spectrum will exhibit significant broadening, making it difficult to accurately locate peak spectral values, thus reducing measurement accuracy. We propose using the K-domain triggering resampling method to address this issue. This method uses the extreme points of the auxiliary path interference signal as resampling points to resample the measurement signal. At this point, the optical frequency intervals between the resampling points are equal. The single-channel distance measurement of the resampled LiDAR can be expressed as R=cτm2n=cτrk′nN′,(7)where τr is the time delay from the calibrated delay fiber, k′ is the peak position after the Fourier transform, and N′ is the total number of sampling points.

The high-precision distance measurement result obtained by averaging the multichannel measurement results of TWDM-FMCW LiDAR can be expressed as R=cnM∑i=1Mfi′τr=cnM∑i=1MKi′Ni′τr,(8)where fi′ represents the beat frequency after frequency resampling for the ith group of signals, Ki′ is the peak position after the Fourier transform of the beat frequency signal for the ith group, and Ni′ is the total number of sampling points for the ith group of signals.

We demonstrate the versatile measurement performance of the TWDM-FMCW LiDAR and its application in 3D imaging using targets such as a cardboard mounted on a linear precision rail, a vertically placed hardboard with hollow patterns, and stair steps.

The multifunctional high-precision measurement performance of the TWDM-FMCW LiDAR uses a cardboard target mounted on a linear precision rail. The experiments were conducted on an optical platform equipped with vibration isolation devices in a controlled environment with constant temperature, humidity, and pressure, aiming to minimize the impact of random phase errors on the measurements. The experimental principle is shown in Fig. 3(a). The tunable laser wavelength range was set to 1545–1555 nm, with a modulation rate of 99.95 nm/s. The balanced detector had a bandwidth of 150 MHz, and the digital oscilloscope signal sampling rate was set to 31.25 MHz. The optical switch array had 5 output channels, each with a switching time of less than 300 ns, matched with a collimator array of 5 channels. To simplify the optical path, a Fizeau common-path interferometer structure was used in the measurement path. A semi-transparent, semi-reflective thin film was coated on the end face of the optical fiber at the output of the optical switch, with both the transmitted and reflected light intensities being 50%. The Fizeau common-path interferometer structure unified the zero points of the distance measurement for each channel at the fiber end face, eliminating system errors. To facilitate the analysis of the accuracy of the proposed FMCW LiDAR distance measurement, the measurement results from the XL-80 interferometer were used as a reference.

The acquired raw single-channel measurement signal is shown in Fig. 3(b). In the left plot, the purple line represents the measurement path signal, and the orange line represents the auxiliary path signal. To eliminate the impact of laser modulation nonlinearity on the measurement, we used the peak and valley points of the auxiliary signal for K-domain triggering and resampling of the measurement signal. The right plot shows the auxiliary sampling points and the resampled measurement signal. During this process, to improve data processing accuracy and the signal-to-noise ratio, we applied a Hanning-windowed FIR high-pass filter to the raw signal. Figure 4 displays the spectrum information of the measurement signal before and after resampling, along with a zoomed-in view of the local spectrum. The blue curve represents the spectrum of the original measurement signal, and the red curve represents the spectrum of the signal after triggering resampling. The spectrum broadening is improved, with the width narrowing, the peak becoming more pronounced, and with lower sideband energy, demonstrating that our method can effectively compensate for the laser frequency modulation nonlinearity.

To evaluate the system accuracy, we performed incremental step measurements and calculated the precision. A piece of hardboard covering multiple channels was used as the target, stably placed on a linear precision rail approximately 1.3 m away from the collimator array. First, we analyzed the overall measurement accuracy of the TWDM-FMCW LiDAR. The target was moved in 20 mm incremental steps, with 10 measurements taken at each step, repeated 8 times, and the average of the five-channel measurements was taken as the result. The measurement results for the entire step movement process are shown in Fig. 5(a). The data characterizes the target’s step motion, with the standard deviation of the 8 sets of measurement results not exceeding 26.86 µm. The mean of each set of measurement data, linear fitting, and residuals are shown in Fig. 5(b). The results indicate good system linearity, with a maximum residual error of 14 µm compared to the interferometer measurements. We continuously measured 10 s at a fixed position, and the results were shown in Fig. 5(c) with the mean value of 1240.556 mm and a standard deviation of 14.73 µm. Figure 5(d) illustrates the Allan deviation of the measurement results, showing a minimum Allan deviation of 189 nm at 2.84 s averaging time. Additionally, we compared the results with traditional FMCW LiDAR distance measurements^[3,5,8,25], as shown in Fig. 5(c). The TWDM-FMCW LiDAR shows less fluctuation in measurement results, demonstrating a significant improvement in ranging precision under this mode.

Next, we analyzed the measurement performance of each channel in the system. Figure 6(a) shows the spectral information and a zoomed-in view of each channel at the same measurement position. Figures 6(b) and 6(c) present the measurement results of each channel at different measurement positions and the deviation from the interferometer measurement values. The measurement precision of each channel remains relatively consistent, with the measurement deviations remaining within 20 µm. This indicates that the system maintains high precision and stability for multichannel distance measurements, demonstrating the excellent performance of the versatile ranging mode.

We demonstrate the application of TWDM-FMCW LiDAR in 3D imaging. To enhance 3D imaging efficiency, we increased the optical switch array and matched the collimator array to 32 channels. Two vertically placed hardboards with “TIF” hollow patterns, spaced approximately 30 cm apart, served as the targets, as shown in Fig. 7(a). A deflection mirror was used as an auxiliary scanning device for horizontal translation. By adjusting the step angle of the deflection mirror, the amount of measurement data was varied. In the experiment, the step angle was set to a fixed value of 0.14°, yielding 64 columns of measurement data. The tunable laser wavelength range was set to 1540–1560 nm, with a modulation speed of 49.98 nm/s. Based on the measurement data, the 32×64 pixels 3D imaging result of the target was reconstructed, as shown in Fig. 7(b). Figures 7(c) and 7(d) show the projection of the imaging point cloud data along the X-axis and Y-axis, respectively, clearly distinguishing the positions of the two layers of hardboard with a measurement precision better than 0.8 cm. Figure 7(e) presents the histogram of the spectral distribution during the scan of the two layers of hardboard, with the spectral distribution matching the positions of the hardboards. The high-precision and clear imaging of the 3D scan results validate the effectiveness of TWDM-FMCW LiDAR in 3D imaging applications.

To further investigate the performance of the LiDAR system in actual 3D imaging applications, we conducted static scene imaging of a corridor staircase. Figure 8(a) shows the experimental setup and scene for static 3D imaging of the staircase. The dashed section indicates the 3D imaging field of view and magnified details. The target scene is 1.8 m away, with the staircase consisting of 11 steps, each step having a depth of approximately 0.3 m. A 2D galvanometer serves as an auxiliary scanner. After the Y-axis completes a scanning cycle by rotating at a specified angle, the X-axis moves to a fixed position to start scanning the next position. The imaging result of the point cloud data processed by software is shown in Fig. 8(b). Figure 8(c) presents the gradient information obtained by the TWDM-FMCW LiDAR scanning the staircase. The 3D imaging results show that the initial position of the staircase is 1.870 m, and each step depth is 0.297 m, consistent with the actual information. This indicates the high precision and reliability of the TWDM-FMCW LiDAR in 3D imaging of actual scenes. Figure 8(d) displays the time-frequency information of each channel within a scanning cycle spanning two steps, reflecting the details of the 3D scanning process of the staircase and the versatile measurement capabilities of the LiDAR system.

This paper proposes a multichannel TWDM-FMCW LiDAR system integrated with the optical switch. This system is characterized by its high measurement speed, high resolution, good stability, and simple structure, making it suitable for multifunctional high-precision measurements and 3D imaging. By implementing TWDM technology for FMCW lasers, the system achieves unified transmission of multiple-length information through a single fiber channel. The TWDM-FMCW LiDAR system can improve measurement accuracy by averaging measurements from multiple channels for the same target. It can also achieve multichannel distance measurements, addressing the needs for anti-interference, non-cooperative, multifunctional, and low-cost precision measurements. In the demonstration experiment of measuring a cardboard target on a linear precision rail, the proposed FMCW LiDAR system achieved an overall absolute distance measurement accuracy better than 14 µm, with each channel achieving better than 20 µm. The overall standard deviation reached 14.73 µm, and the minimum Allan deviation was 189 nm at a 2.84 s averaging time. Experimental results show that the TWDM-FMCW LiDAR system exhibits strong stability in multifunctional distance measurement applications, achieving high-precision and high-accuracy measurements, significantly enhancing efficiency.

Furthermore, by combining the TWDM-FMCW LiDAR with auxiliary scanning devices, it is possible to perform 3D imaging of actual scenes. We demonstrated imaging experiments with “TIF” patterned cardboard and corridor stairs, achieving 3D imaging results that are both reliable and highly precise, with point cloud precision less than 0.8 cm. In summary, the proposed TWDM-FMCW LiDAR offers a simple, efficient, high-precision, and multifunctional solution for distance measurement and 3D target analysis. When integrated with chip-level arrayed waveguide gratings and lens-assisted beam steering technology, the system can be further miniaturized, expanding its potential for practical applications such as precision contour measurement, large equipment assembly, and space exploration^[26–28].