Depth estimation in SPAD-based LIDAR sensors

Mingzhe Chen,Padmakumar Rao,Esteban Venialgo,Padmakumar R. Rao

DOI: https://doi.org/10.1364/oe.507975

IF: 3.8

2024-01-18

Optics Express

Abstract:Mingzhe Chen, Padmakumar R. Rao, Esteban Venialgo In direct time-of-flight (D-TOF) light detection and ranging (LIDAR), accuracy and full-scale range (FSR) are the main performance ... [Opt. Express 32, 3006-3030 (2024)]

optics

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly focus on the accuracy of depth estimation and the full - scale range (FSR) optimization in single - photon avalanche diode (SPAD) - based light detection and ranging (LIDAR) systems. Specifically, the paper focuses on the following aspects:

1. **Selection and evaluation of depth estimation methods**: The paper explores the impact of different depth estimation methods on LIDAR performance, and pays special attention to the effect of artificial intelligence (AI) - based methods in compensating for the non - linearity of depth estimation. This involves how to improve the accuracy and reliability of depth measurement through algorithms.

2. **Trade - off analysis between system parameters and application constraints**: Through simulation and experiment, the paper studies the interdependence between system parameters (such as total optical power \(P_s\), number of time - correlated single - photon counting cycles \(N_{\text{TCSPC}}\), photon detection probability \(\text{PDP}\), and minimum time interval of time - to - digital converter \(\Delta T_{\text{TDC}}\)) and application constraints (such as background light intensity \(E_n\) and target reflectivity \(\rho\)) and their impact on depth estimation performance.

3. **Derivation and application of Cramér–Rao lower bound (CRLB)**: The paper derives an analytical model to calculate the Cramér–Rao lower bound (CRLB) of the LIDAR system, which takes into account the influence of shot noise. CRLB is used to evaluate the statistical efficiency of different depth estimation methods and provides a theoretical performance upper limit.

4. **Optimization of depth range and depth resolution**: Through the adjustment of system parameters, the paper studies how to maximize the depth range and improve the depth resolution under different application conditions (such as smartphone applications).

### Main contributions
- **Improvement of depth estimation methods**: A depth estimation method based on artificial neural network (ANN) is proposed and evaluated. This method can effectively compensate for the non - linear error in depth estimation, thereby improving the accuracy of depth measurement.
- **Comprehensive evaluation of system performance**: Through simulation and experiment, the system performance under different parameter settings is comprehensively evaluated, providing guidance for system design and optimization.
- **Establishment of theoretical model**: A Cramér–Rao lower bound model considering the influence of shot noise is established, providing a theoretical basis for the performance evaluation of depth estimation methods.

### Application background
The paper pays special attention to smartphone LIDAR applications. These applications require high - precision depth measurement under limited resources and strict performance indicators. Therefore, the research results of this paper are of great significance for promoting the development of smartphone LIDAR technology.

### Conclusion
Through theoretical analysis, simulation and experiment, the paper systematically studies the depth estimation problem in SPAD - based LIDAR systems, proposes effective solutions, and provides valuable references for future research and applications.