Trajectory data is a valuable source of information for transportation research and planning. It can be used to study traffic patterns, identify bottlenecks, and improve traffic safety. In recent years, there has been a growing interest in using LiDAR and camera (computer vision) to collect traffic trajectory data. These two technologies have different strengths and weaknesses, and the best choice for a particular application will depend on the specific requirements.
LiDAR (Light Detection and Ranging) is a remote sensing technology that uses laser beams to measure the distance to objects. This technology utilizes the time difference between the transmission of laser beams from the transmitter and their reflection back to the receiver. This unique capability enables LiDAR to create detailed three-dimensional (3D) models of various objects. By undergoing processes such as background filtering, clustering, and referencing, LiDAR can effectively classify and detect road users.
Cameras have been a valuable tool in the transportation industry, initially used to capture evidence for traffic violation cases. In the past, researchers manually tracked road users through frame-by-frame video analysis, but this method was time-consuming and prone to errors. This specialized field of artificial intelligence extracts information from digital images and videos, road users can be automatically identified and tracked using algorithms with advancements in computer vision technology.
The paper titled “Evaluation of Roadside LiDAR-Based and Vision-Based Multi-Model All-Traffic Trajectory Data” explores the use of two different data collection methods [1]. In their study, the authors focused on general location accuracy, volume counting accuracy, detection range, and speed. The data collection was carried out during both daylight and nighttime hours at the same intersection in Nevada, US and over the same time period to ensure comprehensive analysis. Both datasets were recorded at a frequency of 10 Hz to capture detailed information. The accuracy of the data was assessed by comparing it with ground truth trajectories, providing valuable insights into the effectiveness and reliability of both data collection methods.
The study found that LiDAR-based trajectory data has a broader detection range and is less affected by poor lighting conditions than vision-based data. Both sensors performed well for volume counting during daylight hours, but LiDAR-based data was slightly more accurate at night, particularly for pedestrian counting. LiDAR was able to detect pedestrians before they entered the intersection, whereas the camera-based trajectories showed shorter lengths. After applying smoothing techniques, both sensors accurately measured vehicle speeds, but vision-based data showed greater fluctuations in pedestrian speed measurements.
The LiDAR system offers significant advantages, particularly in terms of its extensive detection range and robustness in various lighting conditions. These qualities make it a highly favourable choice for applications that require precise and consistent trajectory data. LiDAR technology has made remarkable strides in multiple industries, finding applications in transportation, environmental monitoring, urban planning, and even the development of autonomous vehicles. Its ability to create detailed 3D models and its efficiency in data processing have revolutionized various aspects of modern society.
On the other hand, computer vision continues to evolve and holds tremendous potential to reshape transportation also other industries through real-time data analysis and decision-making. As advancements in computer vision technology continue, we can expect even greater breakthroughs in efficiency and accuracy. Although computer vision’s accuracy is influenced by video quality, it provides a more cost-effective option for data collection when compared to LiDAR-based methods. Moreover, video cameras are relatively easier to install and maintain compared to LiDAR systems, making them a practical choice for certain applications.
However, it’s essential to consider the data storage aspect. Videos require significantly more storage space for the same time period compared to LiDAR data. This factor should be carefully weighed when deciding on the appropriate data collection method based on specific project requirements.
In conclusion, both LiDAR and computer vision offer valuable contributions to the transportation field. While LiDAR excels in certain aspects such as detection range and lighting conditions, computer vision provides cost-effectiveness and ease of maintenance. As technology advances, the choice between these two methods may depend on the specific needs of each project, keeping in mind factors like accuracy, efficiency, and storage requirements.
Source:
| [1] | Guan F., Xu H., and Tian Y., “Evaluation of Roadside LiDAR-Based and Vision-Based Multi-Model All-Traffic Trajectory Data,” Sensors, vol. 23, no. 12, 2023. |
