From Lab Curiosity to Pocket-Sized Corner Detection Technology
Non-line-of-sight (NLOS) imaging, often described as technology that lets you “see around corners,” has long been confined to research labs. Traditional setups relied on powerful, expensive lasers and highly specialized optics, keeping them far from everyday devices. MIT Media Lab researchers have now shown that consumer-grade hardware, similar to the iPhone LiDAR sensor, can achieve a version of this feat. Their system can detect and track objects that are completely outside the camera’s field of view, using low-power LiDAR rather than bulky lab equipment. The key shift is that this is no longer a purely theoretical demonstration. It runs on hardware that costs under USD 50 (approx. RM230) and uses code the team has publicly released. That combination of affordability and transparency sets the stage for smartphone makers, app developers, and robotics companies to experiment with corner detection technology in real-world products.
How iPhone LiDAR Can Infer What’s Around a Corner
The MIT team’s approach transforms the iPhone LiDAR sensor into a kind of optical echolocation system. Instead of trying to capture a sharp image of what’s hidden, their method focuses on how light subtly bounces from objects around a corner and returns to the sensor. As you move the device, the system continuously records noisy, partial reflections. Using an aperture sampling model, it stitches these imperfect snapshots together over time. The result is not a photo-realistic picture but a progressively richer inference: it can tell that something is there, estimate how it’s moving, and roughly reconstruct its shape. Crucially, it also tracks the camera’s own motion, enabling it to relate those hidden objects to the device’s position. This interplay between motion, LiDAR capabilities, and clever software is what allows phones and tablets to begin to “see around corners” without any exotic new hardware.
From Single Objects to Self-Localizing Robots
In demonstrations, the researchers showed four core abilities built on top of this corner detection technology. First, the system can track a single hidden object, following how it moves even when it never enters the camera’s direct line of sight. Second, it can reconstruct that object’s approximate shape from the LiDAR data. Third, it scales up to multiple objects, distinguishing and tracking several hidden movers at once. The fourth capability is especially significant for robotics: camera self-localization using hidden landmarks. Here, a robot or autonomous device can orient itself by referencing objects it cannot directly see. That means better navigation in cluttered, dynamic environments where line-of-sight is frequently blocked. For delivery robots, drones, or future home assistants, this could dramatically improve obstacle avoidance and path planning, using the same class of LiDAR sensors already appearing in consumer smartphones and tablets.
What This Could Mean for Future iPhones and Everyday Safety
Current high-end iPhones already ship with LiDAR hardware, which is primarily used today for portrait effects, low-light autofocus, and basic augmented reality. MIT’s work suggests these same sensors could eventually power safety and navigation features that work beyond line of sight. Imagine your phone quietly warning you about a cyclist rushing toward an intersection you’re about to cross, or helping you navigate a smoky hallway by tracking hidden obstacles. In augmented reality, apps could anchor virtual elements to objects that are temporarily out of view, making AR scenes more stable and immersive. For emergency responders, phones and tablets might assist in locating moving people behind debris or around corners. While you can’t install this exact MIT system on your iPhone yet—raw LiDAR data access remains locked down—its success shows that the hardware in your pocket is already capable of much more.