Ogier Electronics
Home Products About Us News Knowledge Hub Our Partners Contact Us
Site Security Scan-360 Scan-Sector Scan-Integrated
Perimeter Security Scan-Beam Scan-Sector Super-Beam 200 Super-Beam 400
Stopped Vehicle Detection Track to Train CCTV Broadband Wireless Vehicle & Mobile CCTV CCTV Transmission Homeland Security Systems Custom Solutions
   Close products menu

Comparison of Stopped Vehicle Detection (SVD) Technologies for Smart Motorway Applications

Smart motorways implement real-time monitoring, adaptive speed limits and dynamic lane closures. Schemes such as all-lane running re-use hard shoulders as live traffic lanes. Technological solutions must be used to rapidly detect stationary and broken down vehicles to prevent collisions and to protect all road users.

Ogier Electronics has developed a scanning microwave radar system for SVD.
We believe that radar is the most suitable detection technology available for stopped vehicle detection.

smart motorway

Overview of Various Technologies

Video Analytics

Video analytics uses computer software to analyse footage from video cameras to try to determine what objects the camera can see.
Since cameras use visible light, their range is limited during poor weather such as fog, rain, mist and snow. This requires multiple cameras positioned along the road to cope with the reduced range.

foggy road

Analytics does not work well at night. Although street lighting can be used to mitigate this problem, it adds to the cost of the system and on-going operating costs, as it cannot be switched off during light traffic periods in the middle of the night.

Bright sunlight is a major problem as it dramatically reduces the sensitivity of video cameras, effectively blinding the camera if it is facing the sun. This is a big problem when the rising or setting sun is low in the sky. To mitigate this problem additional cameras are needed so they can point in opposite directions. This further adds to the costs.
For analytics to work reliably, the image must have sufficient contrast and clarity.

video analytics

Cameras are unable to see through smoke, do not function correctly if the lens is dirty and must be orientated correctly to provide overlapping fields of view.

Video analytics uses software routines to analyse the camera pixels to classify the image. These software routines typically utilise machine-learning algorithms where their robustness is hard to prove.
For example Google [1] found that their image classification software could easily be tricked into misidentifying pictures when tiny amounts of pixels (0.7%) were modified. An example is shown below:
AI misidentification

Summary of Video Analytics

  • Large numbers of cameras are required to implement a sufficiently robust and reliable system.
  • Correct illumination is essential.
  • Video analytics software can misidentify images unless conditions are perfect.
  • Image quality must be higher than that needed for human operator.
  • Degraded by smoke, mist, fog, rain, snow, hail, heat haze, bright sunshine, lack of light and dirty camera lens.

Thermal Cameras

Thermal cameras detect temperature differences between objects in the field of view, so see heat rather than light.

thermal camera view

Thermal cameras overcome some of the limitations of optical cameras, such as the requirement for the road to be illuminated correctly.
However they cost more than optical cameras and still have inherent risks with analytic software.

Fog, rain, and high humidity can severely limit the range of thermal imaging systems due to scattering of light off droplets of water.
Thermal cameras additionally degrade in the presence of smoke, fire, hot gasses and heat haze.
Performance also depends on the ambient temperature.

Summary of Thermal Cameras

  • Thermal cameras have many similar problems to optical video cameras.
  • Thermal sensor is more expensive than optical sensor.
  • A considerable number of cameras would still be required for sufficient coverage.

Traffic Counters and Induction Loops

Traffic counters and loops function by counting the number of vehicles that pass a fixed point on the road.
They provide a way to estimate traffic density, perform basic traffic survey counts and to detect vehicles waiting for traffic lights at fixed stopping points.
Various techniques are used such as axle sensors detecting wheels passing a pair of cables, buried inductive loops that act like a metal detector measuring the vehicle as it passes overhead, or (non-scanning) side-fire radars that simply look across a narrow slice of the road counting the number of vehicles that pass by.

traffic counter

Summary of Traffic Counters and Induction Loops

  • These types of system are grossly unsuitable for SVD because they cannot identify the position of a stationary vehicle unless the vehicle happens to stop exactly where the sensor is positioned.

Lidar

Lidar uses a laser scanner to very precisely and rapidly scan an environment to build a detailed three-dimensional image. Laser scanners have very precise imaging capabilities, can measure distance and operate at night.

Lidars share many problems with optical systems where the range is dramatically reduced due to smoke, fog, mist, rain and snow.
The sensor must be kept clean as the laser beam can easily be blocked by dirt over the lens.

Lidar sensors are much more expensive than optical cameras and the field of view tends to be more limited, especially lidars with longer operating ranges which results in blind spots where the sensor cannot see.
laser scan using Lidar

Summary of Lidar

  • Lidar is fundamentally limited to very short ranges because it cannot operate in all weather conditions.
  • Multiple lidar sensors are required to avoid blind spots, which makes them uneconomical to use.

Microwave Radar Advantages

We strongly believe that scanning microwave radar systems overcome many of the limitations of other technologies.

Radar uses short wavelength radio waves (microwaves) to measure the environment. Radar actively transmits a microwave signal and measures the response from every object within the field of view. Radar directly measures the direction, speed, range and size of the objects it detects and operates in all weather conditions.
Radar works regardless of ambient light levels, so works well in complete darkness, in bright sunshine and even in conditions that would dazzle optical sensors.

Microwave radar is not seriously affected by fog, mist, rain or snow. The all-weather capability of radar has been demonstrated for decades by aircraft landing safely in zero visibility conditions, by using radar altimeter, air-traffic control radar and landing approach radar systems.

air traffic control radar

Radar is not affected by smoke, hot gases, heat haze or fires. The radar emits and receives short wavelength microwaves so does not rely on temperature measurements in the same way as thermal cameras and is thus immune to thermal camera deficiencies.

Microwaves readily pass through non-metallic materials so any accumulation of dirt or grime over a radar sensor has minimal effect on the sensor operation, reducing on-going maintenance.

Radar systems for SVD do not require complicated and risky video analytic style software to identify targets. The signal processing is based on techniques that have been developed and proven by maritime, aviation and defence industries over many decades.

radar signals from textbook

Compared to other technologies, radar has a long operating range, resulting in fewer sensors per kilometre. Typical radar spacing can be as high as 500m, which is generally as far apart as is possible anyway due to road curvature and availability of direct line of sight.

A single fixed install position covers a large road area with up to 360-degree coverage around a sensor. Our innovative design also has a very large vertical (elevation) field of view and we have a demonstrated ability to detect vehicles almost directly underneath the radar sensor.


References:
[1] Explaining and Harnessing Adversarial Examples - Ian J. Goodfellow, Jonathon Shlens, Christian Szegedy.

     xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

  © Ogier Electronics 1993-2020  |  Tel: +44 (0)1727 853521  |  Privacy Policy  |  YouTube  |  Last updated: 07:26:34 - Tuesday 28th April 2020