In the field of robotics engineering, tools like Lidar are essential for environmental perception, a fundamental requirement to ensure the autonomy and safety of systems and machines being automated.
Among the most important and widely adopted technologies for three-dimensional mapping and spatial recognition, LiDAR (Light Detection and Ranging) has established itself as an essential tool for navigation.
What is LiDAR
LiDAR is a remote sensing technology that uses laser pulses to measure the distance between the sensor and surrounding objects. It is based on a simple yet highly effective principle: Time of Flight (ToF).
The device emits a laser pulse and measures the time it takes to reflect off an object and return to the receiver.
Since the propagation speed of the laser signal is known (equal to the speed of light), the distance traveled can be calculated by multiplying this speed by the measured time. As the laser travels the path twice (to the object and back), the actual distance between the sensor and the object is half the total path length.
From this operational principle, different types of LiDAR systems can be derived, varying in design, precision, and cost:
- Mechanical LiDAR: Utilizes moving mechanical components (such as rotating mirrors or the rotating sensor itself) to change the direction of the laser beam.
- Solid-State LiDAR: Uses an array of photo-emitters and photodetectors that synchronously emit pulses in different directions in a very short time.
- Flash LiDAR: Emits a single light pulse over a wide field of view and simultaneously measures the time of flight across all sensor photodetectors.
LiDAR systems are also categorized as:
- 2D LiDAR: Emit and receive laser beams on a single plane, usually horizontal.
- 3D LiDAR: Emit and receive laser beams on multiple planes or in various directions.
This is achieved through vertical rotation of the scanning plane or by using multiple overlapping laser planes, known as multi-beam configurations.
How It Works
A LiDAR system typically consists of three main components:
- Laser: Emits high-frequency pulses (up to hundreds of thousands per second).
- Scanner or Rotating Mirrors: Direct the laser beam across a wide field of view.
- Optical Receiver and Processing Unit: Detect the return pulse and compute the distance.
The result is a 3D point cloud that accurately represents the surrounding environment.
Most LiDAR systems also provide a reflectivity index for each point, indicating the surface’s ability to reflect the laser beam. Higher values indicate more reflective surfaces. This property enables the detection and spatial tracking of reflective markers, which can assist with robot localization.
Key Characteristics of LiDAR Sensors
- Range: Often specified for surfaces with 10% reflectivity (low reflectance)
- Precision
- Accuracy
- Point Density
- Scan Frequency
For further technical details:
https://www.sick.com/media/docs/3/63/963/whitepaper_lidar_en_im0079963.pdf
Compared to cameras, LiDAR sensors offer direct and precise distance measurements, covering much longer ranges and being less affected by lighting conditions and visual properties of surfaces.
While cameras provide visually rich information (such as color and texture), they often struggle to estimate depth accurately in the presence of uniform surfaces, low lighting, or backlighting.
LiDAR systems, on the other hand, maintain high performance in darkness or poorly structured environments, making them more reliable for 3D mapping and autonomous navigation. However, unlike cameras, LiDARs do not capture visual information and are generally more expensive and bulkier, which limits their use in low-cost or compact platforms.
For this reason, it is common to combine LiDAR and camera systems, leveraging their complementary features.
LiDAR Applications in Robotics
1. Autonomous Vehicles
Mobile robots, including drones and self-driving vehicles, use LiDAR for obstacle avoidance.
2. Mapping and SLAM
LiDAR is extensively used in SLAM (Simultaneous Localization and Mapping) algorithms, enabling robots to build maps and localize themselves in real time.
3. Warehouse Inventory
In automated warehouses, AGVs (Automated Guided Vehicles) and AMRs (Autonomous Mobile Robots) use LiDAR to navigate complex environments without the need for physical guides.
4. Precision Agriculture
Autonomous tractors and agricultural robots use LiDAR to map terrain, detect surrounding objects, and monitor crops.
Our Experience with LiDAR
Having integrated a wide range of LiDARs over the years on various types of autonomous machines, we’d like to share a few tips that you might not have heard before.
When it comes to highly transparent surfaces—like glass doors, for instance—you need to be cautious about relying on LiDAR measurements. Transparent materials often allow most of the laser beams to pass through without reflecting them back to the sensor, resulting in no returns. In practice, this means the LiDAR may not “see” the glass door at all, and the robot could potentially crash into it.
At Aitronik, we’ve found that integrating ultrasonic sensors is an effective way to reliably detect transparent surfaces and ensure safe robot navigation.
Similarly, for black or non-reflective surfaces, LiDAR can become unreliable for obstacle detection. We’ve encountered this issue multiple times while flying autonomous drones indoors. Although the LiDAR accurately reconstructed the 3D environment with numerous objects, black leather chairs were not detected at all. While this did not pose a safety risk, missing some chairs led to incomplete environmental perception.
We addressed the issue by integrating a small RGB camera, which helped us overcome the LiDAR’s limitations and provided a more complete understanding of the scene. When collecting large volumes of data from LiDARs, it’s crucial to filter and extract only the relevant information for your specific objectives. It’s common practice to process only regions of interest in the 3D point cloud—such as the area in front of the robot—rather than applying algorithms to the full point cloud, including points behind the platform. This approach significantly reduces computational load and ensures high performance for critical runtime algorithms.
Conclusion
LiDAR has become a key technology in modern robotics. Its capabilities are crucial for the safety of autonomous vehicles through collision detection, and it is widely used in both indoor and outdoor localization systems as well as in any robotic application where environmental mapping is required.
