CHAPTER 4. MICROWAVE RADAR SENSORS

INTRODUCTION

Microwave radar sensors gained popularity before and during World War II for object detection. Radar sensors are non-intrusive equipment that transmits electromagnetic signals and receives echoes or energy reflected from targeted objects, including motorized traffic and bicycles on dedicated bicycle lanes. Figure 27 illustrates the operation of a microwave radar to detect moving vehicles. An overhead-mounted microwave radar directs energy toward the approach area, and as a vehicle crosses this projected zone, part of the transmitted energy reflects to the radar’s antenna. The reflected energy is then processed by a receiver, which detects vehicles and calculates traffic flow attributes such as volume, vehicle length, and speed (Klein et al., 2006). Figure 28 and Figure 29 show radar sensors mounted on a traffic signal mast arm and a traffic signal pole, respectively.

Microwave radar sensors used for vehicle presence detection can be categorized into two main types: (a) continuous wave (CW) Doppler waveforms, and (b) frequency-modulated continuous waves (FMCWs). The CW Doppler radar can only detect moving vehicles, while the FMCW radar can detect both moving and stopped vehicles. These sensors transmit energy and receive the portion scattered back into their aperture. The CW Doppler radar is based on the Doppler principle to measure vehicle speed using time-constant frequency signals. The presence of a frequency shift denotes vehicle counts. In contrast, FMCW radar emits a signal where the transmitted frequency continuously varies over time. This variation allows the radar to measure not only the presence and speed of objects but also their distance by analyzing the frequency shift of the returned signal. FMCW radar sensors have various applications in signalized intersection control, including detecting wrong-way vehicles, identifying highway incidents, and providing advance warning to drivers.

Some microwave radar sensors (e.g., Wavetronix SmartSensor) have been used as vehicle presence sensors at signalized intersections for signal control. Traffic count data can be obtained by post-processing the data collected from these sensors. Radar-based systems may require a second sensor upstream from the stop bar for advance detection. Radar sensors are directional and can be configured to monitor approaching or departing traffic. When they are set to target approaching vehicles, departing vehicles are excluded, and vice versa. The radar detectors require vehicles to be moving within their detection zone to be detected. The microwave sensors can maintain a detection signal as long as vehicles move at a minimum speed of 5 mph within the detection zone.

STRENGTHS AND WEAKNESSES

Table 4 summarizes the main strengths and weaknesses of microwave radar sensors for counting motorized traffic.

Table 4. Strengths and Weaknesses of Microwave Radar Sensors.

The validation results from NCHRP Project 03-144 revealed that the accuracy of motorized traffic volumes obtained from radar sensors varied (WMAPE = 1.8% − 31.5%) by location. In general, radar sensors tend to undercount (Figure 30) as traffic volumes increase, particularly at large intersections and locations with a high percent of large vehicles (Chang et al., 2017; Saito et al., 2015; Chamberlin and Fayyaz, 2019).

The most common causes of undercounting are:

• Occlusion: Similar to video-based systems, radar sensors can miss vehicles if the presence of larger vehicles prevents smaller ones from being detected. 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). The problem is more pronounced when sensors are not installed high enough, particularly in large intersections.
• Limited field of view: If the sensor’s field of view is not wide enough to cover all lanes or parts of the intersection, it may miss some vehicles.
• Improper calibration: Incorrectly calibrated sensors may not detect all vehicles accurately.
• Incorrect alignment: Incorrect alignment of the sensor can result in missed detections.
• Detection range: Vehicles too far from the radar sensor may not be detected, especially in large intersections.
• Stop-and-go traffic: Vehicles that stop and start frequently within the detection zone of a sensor may not be consistently detected.

In general, radar sensors are more prone to undercounting than overcounting. However, overcounting can still occur, primarily due to the following factors:

• Cross-traffic interference: The radar sensor might detect vehicles from other approaches or turning lanes.
• Multipath reflections: Radar signals reflecting off nearby objects can cause the sensor to count a single vehicle multiple times.

• Adjacent lane interference: Vehicles in adjacent lanes can also be incorrectly counted, especially if lanes are narrow or the sensor angle is wide.
• Large vehicles: Large vehicles, such as trucks, buses, or pick-up trucks with trailers, may be detected multiple times since they occupy more space and reflect more radar signals than smaller vehicles.
• Shadowing: Large vehicles can create a radar shadow that might cause the sensor to detect the same vehicle multiple times as it moves.
• Sensor placement: Improper placement, such as the radar sensor being too low or at an angle that captures more than the intended lane, can result in overcounting.
• Incorrect calibration: Poor calibration or incorrect configuration of the radar sensor can lead to inaccurate detections and counting.
• Lane splitting: When vehicles move out of their lanes, especially motorcycles or bicycles, the radar might detect them in multiple lanes, leading to overcounting.
• Construction zone interference: Construction-related equipment, signs, or barriers near radar sensors can produce false reflections, contributing to overcounting errors.

RECOMMENDED PRACTICES

Recommended practices and ideal characteristics of microwave radar sensors and data for traffic monitoring use are described below.

Installation and Calibration

• Mounting height: Mount sensors according to manufacturer specifications at a height that allows a clear line of sight over all target lanes without interference. For example, some vendors recommend a mounting height of 14 ft to 20 ft, whereas others suggest 20 ± 5 ft, with a minimum and maximum recommended mounting height of 12 ft and 60 ft, respectively. For large, high-volume intersections, carefully consider placement to reduce occlusion and double-tracking issues since radar accuracy can decrease with higher volumes of large vehicles (Chang et al., 2017; Saito et al., 2015; Chamberlin and Fayyaz, 2019).
• Tilt angle optimization: Adjust the tilt angle of the radar sensors optimally to maximize their detection range and ensure accurate detection across all target lanes. Proper tilting ensures that the sensors can track vehicles over the entire detection area, reducing the chances of missed detections or misclassifications. Some vendors recommend increasing the tilt angle by −2 to −3 degrees for bicycle detection.
• Clear line of sight: Position sensors to avoid obstructions such as buildings, signs, and large trees that can block the radar signal.
• Object interference reduction: Place sensors in locations or adjust sensor settings to avoid interference from building reflections or nearby metallic objects, which can cause false readings or overcounts, particularly in urban settings.
• Signal interference avoidance: Place sensors at strategic locations to prevent interference caused by the signal of other sensors. Signal overlaps can lead to inaccurate detections or confusion between vehicles in different lanes or directions.
• Detection zones: Establish detection zones closer to the radar and outside typical queuing areas to enhance counting accuracy. Smaller zones are recommended to differentiate between closely spaced vehicles, which is especially useful for high-density traffic conditions.

• Multiple sensor placement: Use multiple sensors in complex intersections or where multiple approaches need to be monitored. Each sensor should target a specific approach, minimizing the risk of missed detections and enhancing data accuracy.
• Accurate calibration: Calibrate the sensors accurately during installation to ensure correct vehicle count and classification. Calibration ensures the sensors are aligned properly and can detect vehicles across the desired lanes. A few days of initial monitoring during peak and off-peak times can help validate system accuracy under varying conditions.
• Field of view adjustment: Use sensors with adjustable field of view settings to tailor the detection area to the layout of each intersection. Such settings allow accurate tracking of the intended traffic lanes while avoiding detection in irrelevant areas.
• Multimodal detection: Adjust sensor settings, if available, to enable multimodal detection, especially for intersections with high bicycle traffic volumes. Some radar units allow mode-specific calibration for enhanced accuracy in detecting non-motorized traffic alongside vehicles.

Maintenance

• Routine inspections: Conduct regular inspections to check for sensor misalignment, which can occur due to environmental factors like strong winds or accidental contact with the sensor mounting.
• Cleaning and debris removal: Clean sensors to remove any debris or build-up that could obstruct the signal.
• Automated calibration: Consider using radar sensors with automated calibration routines to maintain accuracy over time. These systems can adjust for minor changes in sensor alignment or environmental factors without manual intervention, ensuring reliable long-term data collection.
• Accessible location for maintenance: Design sensors for easy access and regular maintenance to ensure continuous, accurate performance.