| LiDAR | Measures distance by emitting laser light and detecting its return. Scanning systems collect distance readings across multiple directions. | Can support room mapping, localization, wall following, and route planning. | Provides direct distance measurements and can operate without visible-light image detail. | Performance depends on sensor design and scene conditions. Glass, highly reflective surfaces, occlusion, and objects outside the scan plane can complicate interpretation. |
| RGB-D Camera | Combines a color (RGB) image with depth information. Depth may be obtained using stereo imaging, structured light, or another depth-sensing method. | Can provide visual features and three-dimensional scene information for mapping, object recognition, and obstacle assessment. | Color and depth together can help distinguish scene appearance from distance and shape. | Depth quality and usable range vary by camera technology. Image-based sensing can be affected by lighting, texture, occlusion, and reflective or transparent surfaces. |
| ToF (Time of Flight) | Estimates distance from the time taken for emitted light to travel to a surface and return. Many depth cameras use modulated light and measure phase shift. | Can supply depth data for nearby obstacle detection, distance estimation, and local navigation. | Produces depth measurements across a field of view, rather than only at a single point. | Measurements may be affected by ambient light, multipath reflections, surface reflectivity, and the sensor’s operating range. |
| Obstacle Detection | A perception function—not a single sensor type—that uses inputs such as cameras, depth sensors, LiDAR, bump sensors, or combinations of these. | Identifies or estimates hazards and informs behaviors such as slowing down, changing course, or stopping. | Sensor fusion can combine complementary measurements; physical contact sensors can provide a direct indication of collision. | Detection results depend on sensor placement, field of view, algorithms, and the objects being encountered. No single method guarantees detection of every obstacle. |