weiran-star / traffic_monitoring Goto Github PK

Development of Intelligent Multimodal Traffic Monitoring using Radar Sensor at Intersections

Intelligent transportation systems (ITS) significantly change our communities by improving the safety and convenience of people’s daily mobility. The system relies on multimodal traffic monitoring, that needs to provide reliable, efficient and detailed traffic information for traffic safety and planning. Signalized traffic intersections are critical spots for collecting such mix-traffc data because the most conflicts and crash occurrences involve multiple transportation modes, such as pedestrians, bicyclists, motorcyclists, and cars. How to reliably and intelligently monitor intersection traffic with multimodal information is one of the most critical topics in intelligent transportation research.
This project will investigate a low-cost, low-weight, compact size, and reliable monitoring platform. This platform that incorporates mmWave radar and the machine learning technique to collect multimodal traffic data at intersections is robust to light and adverse weather conditions. The products of this project consists of 1) a prototype of the proposed multimodal traffic monitoring platform using mmWave radar, 2) the real-world experimental dataset collected by the platform for multimodal traffic, and 3) a demo platform at a road intersection to illustrate the performance in terms of measuring multimodal traffic counts, speeds, and directions.
To evaluate the claimed advantages of the proposed radar-based method in this project, a comprehensive real-world traffic data during various time periods, e.g., daytime and nighttime, and various weather conditions will be collected. In addition, a dataset will be built for sharing among researchers and an analysis tool will also be developed to help others evaluate the potential benefits of our approach for improving the multimodal traffic monitoring.

The proposed monitoring platform will be able to collect traffic count, speed and classification data on all lanes at the intersections for each traffic mode in real-time, and also create automatic traffic statistical reports for traffic planners for later analysis. In this project, a demonstrable version of multimodal traffic monitoring platform will be developed at an intersection covered with a wide field of view. The demo system as shown in the figure above will be built to demonstrate the performance of the multimodal monitoring in real time.

Please refer to ~/traffic_monitoring/data/README for data collection instructions.
Experiment in a parking lot (refer to ~/traffic_monitoring/data/training/ for the collected data):

Experiment at a traffic intersection (refer to ~/traffic_monitoring/data/data_intersection/ for the collected data):

The outcome of this project has been organized as a conference paper, which has been accpeted by the IEEE International Radar Conference 2020. Please cite this work as:

F. Jin, A. Sengupta, S. Cao and Y. Wu, "MmWave Radar Point Cloud Segmentation using GMM in Multimodal Traffic Monitoring," 2020 IEEE International Radar Conference (RADAR), Washington, DC, USA, 2020, pp. 732-737, doi: 10.1109/RADAR42522.2020.9114662.