🔬 applied physics

Shot Noise Limited Triangulation

This paper presents a system-level architecture that preserves fundamental information through analog signal processing and common-mode noise rejection, achieving nanometer-scale depth precision at a 1.42-meter standoff with a 10-centimeter baseline, representing a significant improvement over existing triangulation systems while approaching the shot noise limit.

Original authors: John C. Howell, Andrew N. Jordan

Published 2026-05-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: John C. Howell, Andrew N. Jordan

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Idea: Measuring Distance with "Super-Senses"

Imagine you are trying to guess how far away a friend is standing in a foggy field. You have two eyes (detectors) separated by a small distance (your baseline). By looking at the angle your eyes need to turn to focus on your friend, you can calculate the distance. This is called triangulation.

Most cameras and sensors today are like people with slightly blurry vision or shaky hands. They lose tiny bits of information when they take a picture, turning a smooth, continuous signal into a jagged, digital one. This "pixelation" and electronic noise make it hard to measure distance with extreme precision, especially when the object is far away.

The authors of this paper asked: What if we could build a system that doesn't lose any of that information? They designed a new way to measure distance that gets as close as physically possible to the "perfect" limit allowed by the laws of physics (specifically, the random nature of light particles, called photons).

The Problem: The "Digital Staircase" vs. The "Smooth Ramp"

Think of a standard camera measuring light like a person trying to measure the height of a hill by counting steps.

Standard Cameras: They take a picture and count pixels. If the light moves just a tiny bit, the camera might not notice until it moves a whole "step" (a pixel). This creates a "digital staircase" where small movements are lost.
The Authors' Approach: They wanted a "smooth ramp." Instead of counting steps, they wanted to measure the exact slope of the hill.

To do this, they avoided converting the light signal into digital numbers too early. Instead, they kept the signal analog (smooth and continuous) for as long as possible, allowing them to detect incredibly tiny shifts that a normal camera would miss.

The Solution: The "Noise-Canceling" Team

The team built a special system with two main parts working together:

The "Rough Finder" (The Camera): A standard camera acts like a scout. It finds the general direction of the light source (the "friend" in the fog). It doesn't need to be super precise; it just needs to point the system in the right ballpark.
The "Precision Team" (Balanced Detectors): Once the camera points the system at the target, two special sensors (called balanced detectors) take over.
- How they work: Imagine two people standing side-by-side holding a scale. If a gust of wind (noise) hits them both, the scale stays balanced because the wind pushes both sides equally. But if one person moves slightly (the signal), the scale tips.
- The Trick: The system measures the difference between the two sensors. This cancels out the "gusts of wind" (common noise like flickering lights or electrical static) and highlights only the tiny movement of the light source.

They even added a second layer of this trick: they compared the difference between the left sensor and the difference between the right sensor. This is like having a team of four people canceling out noise from all directions, leaving only the purest signal.

The Results: Nanometer Precision

The team tested this system with a small light source (an LED) from about 1.4 meters away. The distance between their two sensors was only 10 centimeters (about the width of a hand).

The Achievement: They measured the depth of the light source with nanometer-scale precision. To visualize this: a nanometer is one-billionth of a meter. It is roughly the width of a few atoms.
The Comparison: While they didn't quite reach the absolute theoretical limit (the "shot noise limit," which is the ultimate speed limit of light itself), they were two orders of magnitude (100 times) better than current camera-only systems and significantly better than other specialized detectors.

Why This Matters (According to the Paper)

The paper emphasizes that the novelty isn't in inventing new physics, but in preserving information.

Old Way: Measure light $\rightarrow$ Turn into digital numbers $\rightarrow$ Lose tiny details $\rightarrow$ Calculate distance.
New Way: Measure light $\rightarrow$ Compare analog signals to cancel noise $\rightarrow$ Keep all the tiny details $\rightarrow$ Calculate distance.

They note that while this system is incredibly precise for a single point (like a laser dot or a star), it is currently too complex to map a whole 3D room like a standard camera does. However, for applications where you need to track a single object with extreme accuracy—like guiding a robot arm or tracking a specific star—it represents a massive leap forward.

Summary Metaphor

If measuring distance with a normal camera is like trying to hear a whisper in a noisy room by shouting "Can you hear me?" and guessing the volume, this new system is like putting on high-end noise-canceling headphones that filter out the room noise entirely, allowing you to hear the whisper perfectly clearly. They didn't make the whisper louder; they just stopped the noise from drowning it out.

Technical Summary: Shot Noise Limited Triangulation

Problem Statement
Passive triangulation is a fundamental technique for determining the 3D position of an object using angular disparity from laterally displaced observations. While effective for lateral coordinates, depth precision in triangulation scales quadratically with distance, making it a poor estimator when the object distance significantly exceeds the baseline (the distance between detectors). Existing camera-based stereo systems are typically limited to precision greater than tens of microns due to technical constraints such as digitization noise, signal discretization, and readout noise. These limitations effectively destroy the Fisher information available in the angular estimation stage, preventing systems from approaching fundamental physical limits. While active systems (e.g., Lidar) can achieve nanometer precision, passive systems relying on photon statistics have not been optimized to this degree.

Methodology
The authors propose a system-level architecture designed to preserve fundamental information throughout the measurement pipeline, specifically targeting the shot noise limit for passive triangulation of a quasi-monochromatic point source. The approach diverges from standard digital-first processing by maintaining analog signals and employing common-mode noise rejection.

Theoretical Framework: The paper derives the Cramér-Rao bound for angle sensing applied to triangulation. It establishes that the depth uncertainty ( $\delta z$ ) is fundamentally limited by the angular uncertainty ( $\delta \phi$ ), which is bounded by the shot noise limit: $\delta \phi \gtrsim \lambda / (\sqrt{N}D)$ , where $N$ is the number of photons and $D$ is the aperture diameter. The authors demonstrate that standard digitization prior to differential measurement introduces noise that far exceeds this fundamental bound.
System Architecture: The experimental setup utilizes a monolithic camera/balanced detector system.
- Coarse Localization: A single 12-megapixel camera provides a rough estimate of the transverse coordinates to guide the system.
- Fine Measurement: Two balanced detectors (specifically New Focus 2901 Quadrant Detectors and in-house Hamamatsu segmented photodiodes) are mounted on a rotating platform.
- Signal Processing: The system employs a "doubly-layered" analog voltage differential scheme. First, the analog voltages from the quadrants of each detector are differenced to reject common-mode noise and determine the centroid shift. Second, the differential outputs of the two balanced detectors are differenced again. This second differencing allows for further common-mode noise rejection (e.g., LED power fluctuations) and amplification of the signal.
- Calibration: The system relies on a monolithic platform to ensure fixed yaw angles ( $\alpha_1, \alpha_2$ ) are calibrated beforehand, removing the need for high-accuracy optical encoders during operation. The depth is calculated using the relation $z \approx b / (\alpha_1 + \alpha_2 + \beta_1 + \beta_2)$ , where $\beta$ represents the offset from the optical axis derived from the balanced detector signals.

Key Contributions

System-Level Architecture: The primary contribution is the design of a measurement pipeline that minimizes information loss by keeping signals analog until the final differential stage, contrasting with systems that digitize early and lose Fisher information.
Application of Cramér-Rao Bounds to Triangulation: The authors explicitly apply the well-known Cramér-Rao bound for angle sensing to the specific context of triangulation depth estimation, highlighting why existing systems fail to reach these limits.
Doubly-Differenced Analog Processing: The implementation of a second layer of analog differencing between two balanced detectors to enhance precision and reject common-mode noise.

Experimental Results
The system was tested with a 10 cm baseline and a standoff distance of 1.42 meters, using a 1 $\mu$ W LED source.

Precision: The system achieved nanometer-scale depth precision. Specifically, the measured precision was approximately 1.1 $\mu$ m RMS for a 1 Hz sinusoidal oscillation, with an amplitude spectral density indicating a precision of roughly 13 nm Hz $^{-1/2}$ at 10 meters.
Comparison: While the results are roughly two orders of magnitude above the theoretical shot noise limit (predicted at ~50 pm Hz $^{-1/2}$ for the specific conditions), they represent several orders of magnitude improvement over current camera-only or other position-sensitive detector systems, which are typically limited to tens of microns.
Linearity: The double-differential signal exhibited a steep linear region, allowing for high-accuracy single-shot distance estimates even with a low-angular-accuracy camera pointing system.

Significance and Claims
The paper claims that its novelty lies not in new basic physics, but in the architectural design that preserves fundamental information. The authors state that their Fisher-information preserving system represents a departure from existing methods that treat measurement noise as an external parameter rather than a fundamental constraint.

The results demonstrate that passive triangulation can achieve unprecedented precision for single point sources, comparable in angular sensitivity to the Gaia star surveyor, albeit in a different context. The authors note that while the current system is a proof-of-principle for single-point depth estimation and is not yet suitable for complex 3D mapping of distributed targets, it offers a pathway for high-precision targeting and tracking applications. They suggest that with further mitigation of vibration, turbulence, and temperature fluctuations, and potentially the use of photon-counting cameras, the system could approach the fundamental limits of triangulation. The work also highlights the potential for satellite-based applications where atmospheric turbulence is negligible.