CHAPTER 5. LiDAR SENSORS

INTRODUCTION

LiDAR is a remote sensing technology that measures distances by emitting pulsed laser beams toward a target and analyzing the reflected signals. LiDAR sensors can detect both motorized and non-motorized traffic. A LiDAR system is composed of three elements: a laser that is used to identify objects, a scanner, and a customized GPS receiver. LiDAR sensors can be categorized into flash LiDAR and rotating LiDAR. Flash LiDAR uses a single light pulse to scan a scene, generating high-density laser points within a focused area. In contrast, rotating LiDAR employs rotating assemblies or mirrors to direct laser beams, enabling a 360-degree scan of the surroundings. LiDAR can generate highly accurate 3D models of the surrounding environment, making it a powerful tool for various applications, including traffic data collection. LiDAR can detect vehicles on roadways, measure their speeds, classify their types, and even monitor pedestrian movements, making it a useful technology for traffic management, transportation planning, and other fields.

LiDAR works by rapidly emitting thousands of laser pulses per second, which reflect off objects such as vehicles or infrastructure. The sensor calculates each object’s exact distance and position by analyzing the time delay and wavelength of the returning pulses. This computation allows LiDAR to create real-time, detailed spatial maps of traffic conditions. The main LiDAR features typically used to select a model are the laser channels (i.e., the total number of laser beams), the vertical field of view, and the vertical resolution or angle between adjacent laser beams. Models containing higher laser channels tend to be more expensive but more accurate and provide improved coverage versus cheaper sensors with fewer beams.

The two primary types of LiDAR sensors are topographic and bathymetric. Topographic LiDAR, which uses near-infrared laser light, is commonly employed for ground-based applications, including traffic monitoring. Bathymetric LiDAR, which uses water-penetrating green light, is more specialized for measuring underwater terrains, such as riverbeds or seafloors (Jagirdar et al., 2019). While bathymetric LiDAR is less relevant to traffic data collection, topographic LiDAR has become a valuable tool for transportation agencies, enabling them to control signals, gather detailed traffic counts, monitor traffic flows, and enhance roadway safety by identifying patterns of congestion or incidents in real time.

In recent years, various LiDAR systems have become commercially available, providing more flexible and cost-effective options for traffic signal control and data collection. These systems can be mounted on infrastructure like traffic poles, bridges, or even drones, providing a bird’s-eye view of traffic conditions. Figure 31a and Figure 31b show a LiDAR sensor mounted on a traffic signal pole. LiDAR sensors (Figure 31) typically have a field of view ranging from 120 to 360 degrees, enabling them to monitor multiple directions of an intersection. Their detection range can span from a few feet to several hundred feet, with the typical range for sensors used at signalized intersections being approximately 164–656 ft (50–200 m).

LiDAR sensors perform well in various lighting conditions, including nighttime, as they rely on emitted rather than ambient light. This feature gives them an advantage over traditional optical cameras for traffic monitoring. However, despite its increasing use in data collection, LiDAR’s application for real-time traffic control remains limited due to the need for significant post-processing of data, which can be computationally intensive. This processing typically involves filtering the noise, classifying objects, and integrating data into traffic management systems. As computational power and data processing algorithms advance and as LiDAR costs

continue to decrease, it could play a more significant role in real-time traffic management and smart transportation systems.

STRENGTHS AND WEAKNESSES

Table 5 summarizes the main strengths and weaknesses of LiDAR sensors for counting motorized and non-motorized traffic.

Table 5. Strengths and Weaknesses of LiDAR Sensors.

The validation results from NCHRP Project 03-144 revealed that the counting accuracy (WMAPE) of motorized traffic volumes obtained from four LiDAR sensors was generally high, ranging between 1.5% to 8.4%. The correlations (R) between LiDAR and benchmark volumes were also high (0.99–1.00). However, the accuracy of temporary LiDAR sensors in counting non-motorized traffic was lower, with WMAPE ranging from 44.9% to 54.5%, and R between 0.73 and 0.79. The equipment did not exhibit a consistent undercounting or overcounting trend.

The most common causes of undercounting are:

• Occlusion: Like other non-intrusive sensors, LiDAR sensors can miss objects if larger fixed or moving objects block the sensor’s line of sight. Occlusion is often observed in lanes farthest from the sensor, where the vertical field of view is limited, especially when two or more adjacent lanes allow the same vehicle movement (e.g., two left-turn lanes or two through lanes).
• Low reflectivity: Objects with low reflectivity such as dark-colored cars, clothes, or gear might not reflect enough laser energy back to the sensor, leading to undercounting.
• Improper sensor position: If the sensor is mounted too high, too low, or in an inappropriate position, it may fail to effectively detect all target objects, particularly in complex intersection layouts.

• Incorrect calibration: Incorrectly calibrated sensors may not detect all objects accurately.
• Adverse weather conditions: Adverse weather conditions like heavy rain, fog, or snow can scatter the LiDAR signals, reducing the effective range and accuracy of detection.
• Limited field of view: A limited field of view with blind spots may prevent the sensor from capturing all objects, particularly non-motorized traffic in large intersections.
• Detection range: Vehicles too far from the sensor may not be detected, especially in large intersections.
• Groups of pedestrians: Pedestrians and bicyclists frequently move in groups, making it challenging for sensors to distinguish and accurately count individuals within a group.

Common causes of overcounting are:

• Reflective surfaces: Highly reflective surfaces (e.g., metallic objects and windows) and multiple reflections from the same object can lead to false detections. Rapid changes in lighting conditions, such as headlights from cars, can result in false counts.
• Moving objects: Moving objects such as leaves, trees, trash, flags, and animals may be incorrectly detected as traffic.
• Incorrect object classification: LiDAR sensors sometimes have difficulty distinguishing between different types of objects, leading to overcounting.
• Incorrect calibration: Overly sensitive calibration settings can lead to LiDAR detecting and counting irrelevant objects.

RECOMMENDED PRACTICES

Recommended installation, calibration, and maintenance practices related to LiDAR sensors are described below.

Installation and Placement

• Mounting height: Install sensors at optimal height (typically 16–20 ft) according to manufacturer specifications to reduce occlusion caused by large objects. Common installation points include traffic signal poles, mast arms, and other high-mounted positions. Note that installing the sensor higher than the manufacturer’s recommended height may also result in occlusion due to undetectable areas outside the sensor’s vertical field of view. For example, as shown in Figure 32, two pedestrians remain undetected. Pedestrian 1, located close to the sensor, is missed because the sensor’s lowest laser beam aims above the pedestrian.
• Sensor angle: Angle sensors to ensure they capture the movement of all motorized and non-motorized road users accurately, focusing on critical areas such as stop bars, crosswalks, and merging points.
• Sensor position: Install sensors at strategic positions to avoid blind spots; provide a clear, unobstructed view of the approaches of interest; and minimize interference from reflective surfaces like glass or shiny road signs that can cause ghost signals.
• Number of sensors: Use one or more LiDAR sensors, as recommended by each manufacturer, to capture all motorized and non-motorized traffic at the approaches of interest.
• Secure mounting: Use stable and vibration-resistant mounts to prevent sensor movement due to strong winds and vibrations from traffic.

• Environmental protection: Ensure sensors are protected from environmental factors such as dust, dirt, salt spray, water ingress, and moisture that could affect performance.

Calibration and Maintenance

• Initial calibration: Calibrate sensors after installing them to define detection zones, adjust sensitivity, and optimize filtering settings. Configure sensors to distinguish between vehicles, pedestrians, and bicyclists. Misalignment or incorrect calibration can lead to incorrect data, making frequent calibration checks necessary. Set appropriate detection range to avoid false detection of objects outside the approaches of interest. Configure the sensor to filter out interference from headlights, streetlights, or sudden changes in lighting conditions.
• Initial data validation: Validate count and other types of data recorded by the LiDAR system by comparing them against benchmark data (e.g., manual counts or manually reduced data from videos). Adjust upon validation, as needed.
• Frequent cleaning: Schedule frequent cleaning, particularly in environments prone to dust or moisture. Dirt and debris on the sensor lens can interfere with LiDAR’s ability to capture accurate data.
• Routine inspections: Conduct routine inspections to check for physical damage and identify potential functionality or misalignment issues caused by weather, vandalism, or accidents.
• Field testing: Conduct regular field tests to verify counting accuracy and signal control effectiveness for both motorized and non-motorized traffic.
• Periodic calibration: Perform regular calibration to maintain accuracy and reliability. Over time, even small shifts in sensor placement or environmental changes can affect accuracy.
• Performance monitoring: Use system diagnostics to identify potential data anomalies and correct issues proactively.