Coherent LiDAR systems — of which FMCW (Frequency-Modulated Continuous-Wave) is the principal automotive implementation — rely on heterodyne detection to extract range and velocity simultaneously from a single measurement. In heterodyne detection, the backscattered optical signal is mixed with a local oscillator (LO) derived from the same laser source, generating a beat frequency that is linearly proportional to target range (via the chirp rate) and Doppler-shifted by target radial velocity.

This dual-observable property fundamentally differentiates coherent systems from direct-detection or incoherent ToF systems, which can only extract range through pulse timing. The ranging advantage is demonstrated in Blackmore Sensors & Analytics' patent filings, which explicitly reference "coherent processing to detect Doppler shift" as a core enabling feature, defining an SNR-range relationship that is central to why heterodyne detection enables longer and more precise ranging than direct-detection alternatives.

As documented by WIPO patent filings and confirmed by Aeva Inc.'s 2023 research, FMCW LiDAR provides per-return instantaneous radial velocity measurements that can be used to correct motion distortion in mechanically scanned sensors. Doppler-aided continuous-time odometry outperforms methods that lack this velocity channel, particularly in geometrically degenerate environments where range-only alignment fails.

By contrast, a purely incoherent architecture relying on pseudorandom codes for indirect ToF — as described in research from the National Institute of Telecommunication, Brazil (2021) — achieves range errors of less than 0.6% across 13–1,000 m, but provides no velocity information and relies entirely on code correlation rather than optical coherence for precision. The absence of a local oscillator means the receiver noise floor is determined by shot noise and detector thermal noise without the coherent gain that heterodyne mixing provides, placing fundamental limits on sensitivity and range resolution.

Heterodyne detection enables shot-noise-limited sensitivity, directly improving ranging accuracy by providing coherent gain over direct-detection architectures. This is the foundational mechanism behind the SNR-based scan optimization described in Blackmore's 2020 LIDAR System for Autonomous Vehicle patent.

Simultaneous range and Doppler extraction from a single coherent measurement eliminates the need for multi-frame temporal differencing to obtain target velocity, as validated by real-world FMCW odometry results from Aeva Inc. (2023). The IEEE-published research confirms Doppler-aided continuous-time odometry outperforms range-only methods in degenerate environments.

Adaptive range resolution is a natural capability of frequency-modulated coherent systems, as demonstrated in GM Cruise Holdings LLC's 2024 patent applying finer resolution at close range and coarser resolution at longer range within a single FMCW architecture.

Range aliasing, a dominant accuracy failure mode in pulsed direct-detection LiDAR, requires complex algorithmic measures from Waymo (time-varying dither, multiple hypotheses; extended detection periods) — problems that coherent beat-frequency encoding inherently mitigates at the signal level. The EPO patent record confirms active prosecution of aliasing mitigation across multiple Waymo families.

The broader industry trend, tracked across the PatSnap analytics platform, shows a shift from purely algorithmic fixes toward first-principles coherent system designs as chip-scale photonic integration reduces the cost barrier historically making direct-detection systems the default choice. Incoherent ToF architectures demonstrate competitive range error (<0.6% across 13–1,000 m) but lack the velocity channel and coherent gain of heterodyne systems, confirming the fundamental performance ceiling difference between the two paradigms.