Time of Flight (ToF) is a sensing technique that determines the distance to an object by measuring the elapsed time between the emission of a signal — typically light, sound, or radio waves — and the detection of its reflection. In depth sensing, ToF cameras emit pulsed or modulated infrared light and record the per-pixel round-trip delay to construct a dense depth map. ToF sensors are widely deployed in robotics, autonomous vehicles, augmented reality, and gesture recognition due to their ability to produce real-time depth data without ambient light dependency.
Overview
- ToF depth cameras achieve frame-rate depth sensing (typically 30–60 fps) without requiring structured patterns or stereo baseline, making them compact and suitable for mobile devices.
- Interference from multipath reflections and sunlight saturation are key limitations, particularly outdoors.
- LiDAR systems are a high-fidelity variant of ToF operating over longer ranges with rotating or solid-state emitter arrays.
Key Aspects
- Direct ToF (dToF): Measures round-trip time of a single pulse; used in LiDAR.
- Indirect ToF (iToF): Modulates a continuous wave and measures phase shift; used in depth cameras (e.g., Intel RealSense, Microsoft Azure Kinect).
- Resolution: Typically lower spatial resolution than RGB cameras but provides per-pixel depth.
- Range: Consumer devices cover 0.1–5 m; automotive LiDAR reaches 200+ m.
Mechanisms
- For iToF, the sensor emits a sinusoidally modulated IR signal; the phase difference between emitted and received signals encodes distance.
- For dToF, a picosecond laser pulse is emitted and a SPAD (single-photon avalanche diode) array times the photon return.
- Signal Processing algorithms remove multipath artefacts and apply temperature compensation.
Applications
- Face ID and face unlock on smartphones using structured ToF arrays.
- Obstacle avoidance in warehouse Robotics and drone navigation.
- Hand tracking for Augmented Reality headsets.
- 3D Reconstruction of environments for SLAM-based mapping.