By the end of this module, you will be able to:
- Understand what stereo depth sensing is and how it works
- Compare RealSense cameras with other depth technologies
- Identify the right RealSense camera for your application
- Recognize common use cases and applications
RealSense is a family of 3D depth cameras that use stereo vision to perceive the world in three dimensions. Unlike traditional 2D cameras that only capture color information, RealSense cameras provide both color (RGB) and depth (D) data, creating what's called RGB-D imaging.
RealSense cameras work similarly to human vision:
- Two cameras (like your eyes) capture slightly different views of the same scene
- Software algorithms compare these views to calculate depth
- Depth information is generated by finding corresponding points between the two images
- 3D point cloud is created with X, Y, Z coordinates for each pixel
Left Camera View Right Camera View Depth Map
👁️ 👁️ 📊
[Image] [Image] [Distance]
| Technology | How It Works | Pros | Cons | Best For |
|---|---|---|---|---|
| RealSense (Stereo) | Two cameras + algorithms | Cost-effective, good range, color + depth | Affected by lighting, texture dependent | General robotics, AR/VR |
| LiDAR | Laser pulses + time-of-flight | Very accurate, works in any light | Expensive, no color, large size | Autonomous vehicles, mapping |
| Time-of-Flight (ToF) | Infrared light pulses | Fast, good for close range | Limited range, affected by sunlight | Gesture recognition, close-range |
| AI Cameras | Single camera + AI inference | Low cost, no special hardware | Less accurate, requires training | Consumer applications |
- Best for: Indoor applications, close-range precision
- Depth range: 7cm - 50cm
- Resolution: 1280×720 @ 90fps
- Key features: Global shutter, good for moving objects in close range
- Use cases: Robotics, arm/hand positioning, quality inspection
- Best for: Indoor applications, close-range precision
- Depth range: 0.3m - 10m
- Resolution: 1920×1080 @ 30fps
- Key features: Global shutter, good for moving objects
- Use cases: Robotics, 3D scanning, quality inspection
- Best for: General-purpose applications
- Depth range: 0.1m - 10m
- Resolution: 1280×720 @ 30fps
- Key features: Most popular, balanced performance
- Use cases: Robotics, AR/VR, computer vision
- Best for: Long-range and outdoor applications
- Depth range: 0.4m - 20m
- Resolution: 1280×720 @ 30fps
- Key features: Extended range, IMU sensor
- Use cases: Autonomous navigation, outdoor robotics
- Best for: AI-optimized applications
- Depth range: 0.1m - 10m
- Resolution: 1920×1080 @ 30fps
- Key features: AI acceleration, GMSL
- Use cases: Edge AI, smart cameras, robotics, industrial
- Best for: High-precision applications
- Depth range: 0.6m - 6m
- Resolution: 1280×800 @ 60fps
- Key features: Power over Ethernet (PoE), ROS2, Holoscan, IP65
- Use cases: 3D scanning, precision measurement, robotics, industrial
- Autonomous navigation: Obstacle detection and path planning
- Manipulation: Object recognition and grasping
- SLAM: Simultaneous Localization and Mapping
- Human-robot interaction: Gesture recognition and safety
- Hand tracking: Natural interaction without controllers
- Body tracking: Full-body motion capture
- Room mapping: Spatial understanding for AR
- Gesture control: Touchless interfaces
- Quality inspection: 3D measurement and defect detection
- Robotic assembly: Precise object manipulation
- Safety systems: Worker protection and monitoring
- Inventory management: Automated counting and tracking
- Surgical navigation: Precise instrument tracking
- Rehabilitation: Movement analysis and therapy
- Patient monitoring: Fall detection and activity tracking
- Medical imaging: 3D scanning and modeling
- Driver assistance: Pedestrian and obstacle detection
- Autonomous driving: Environmental perception
- Interior monitoring: Driver attention and safety
- Parking assistance: Precise distance measurement
| Model | Depth Range | RGB Resolution | Depth Resolution | FOV (H×V) | IMU |
|---|---|---|---|---|---|
| D415 | 0.3-10m | 1920×1080 | 1280×720 | 69°×42° | No |
| D435 | 0.1-10m | 1920×1080 | 1280×720 | 87°×58° | No |
| D455 | 0.4-20m | 1920×1080 | 1280×720 | 87°×58° | Yes |
| D457 | 0.1-10m | 1920×1080 | 1920×1080 | 87°×58° | Yes |
| D555 | 0.4-20m | 1920×1080 | 1024×768 | 70°×55° | Yes |
- D435: Best balance of features and price
- D415: If you need global shutter for moving objects
- D455: Long range and IMU for navigation
- D435: General-purpose robotics applications
- D555: Highest accuracy for measurement
- D415: Good precision with global shutter
- D457: AI acceleration and latest features
- D435: Cost-effective for learning and development
- Think about your intended application
- Research which RealSense camera would be best
- Consider factors like:
- Required depth range
- Lighting conditions
- Accuracy requirements
- Budget constraints
- List 3 applications where RealSense would be better than LiDAR
- List 3 applications where LiDAR would be better than RealSense
- Explain your reasoning for each choice
-
What is the main advantage of stereo depth sensing over single-camera depth estimation?
- A) Lower cost
- B) Higher accuracy
- C) Better color quality
- D) Faster processing
-
Which RealSense camera is best for outdoor applications?
- A) D415
- B) D435
- C) D455
- D) D555
-
What does RGB-D stand for?
- A) Red-Green-Blue-Depth
- B) Real-Good-Best-Depth
- C) Range-Gradient-Brightness-Distance
- D) None of the above
Congratulations! You now understand the fundamentals of RealSense cameras. In the next module, you'll learn how to set up and install the RealSense SDK.
Ready to continue? → Module 2: Setup & Installation