Power lines and other wires pose serious danger to helicopters, especially during low-level flight. The problem with such obstacles is that they are difficult to see until the helicopter is close to them, making them hard to avoid. Even large pylons, towers and tall trees are tricky to spot and avoid if they do not stand out clearly from their backgrounds.
This is a challenge that EADS Deutschland GmbH - Defence Electronics of Germany took on when it launched its HELLAS-Warning System (HELLAS-W) in 2000. The system uses ladar (lasar radar) and other systems to give early warning of obstacles in the flight path. It is currently being used by the German Federal Police and the Royal Thai Air Force. And next year EADS Deutschland will begin integrating the system's successor into military helicopters.
Figure 1. HELLAS-A.
The new HELLAS-Awareness System (HELLAS-A, Figure 1), which is currently at a qualification stage, continually scans the three-dimensional space in front of the helicopter with a laser beam. A short laser pulse is emitted in a known direction. Evaluation of the time elapsed until the arrival of the laser energy (time-of-flight measurement), after it has been scattered by an obstacle like a power line, gives the distance and elevation angle of the obstruction. HELLAS-A overlays this information -- in the form of simple symbols, prioritized in terms of their proximity to the flight path -- over other sensor information, such as FLIR, and database information, such as digital maps (Figure 2).
Figure 2. Obstacle symbology overlay over FLIR video.
The HELLAS-A sensor is designed to detect a 5mm wire in a distance of more than 700m at a visibility of 12 km with a detection probability of 99.5 percent per second. "The algorithms find wires seconds before you would see them with your eyes," said Christian Seidel of EADS Deutschland, when he described this new system at SPIE Europe's recent Security + Defence symposium in Cardiff, UK.
This system can be fully integrated in the helicopter avionic system. Its results can be displayed on the helicopter's multifunctional display (MFD) units or on a pilot's head-mounted sight and displays (HMS/D). "We want to reduce the workload for pilots as much as possible," said Seidel.
HELLAS-A plans to address another common challenge too. When helicopters land on sandy or dusty ground they experience "brownout" where their visibility is blocked by the disturbed sand or dust (Figure 3). Such situations are very disorienting for the pilots and can lead to accidents as reference points are obscured. HELLAS-A tackles this with brownout recovery, a three-dimensional see-and-remember system. It does this using its situation-awareness suite that has a range of different detectors, including radar and ladar.
Figure 3. German CH53 in brownout. Source: Heeresflieger Laupheim.
"It keeps in memory what is detected and shows this when you get a brownout situation," explained Seidel.
There is still plenty more research to be done in this area though. The ultimate goal of the HELLAS systems is to enable helicopters to operate 24 hours a day, seven days a week. This means flying in zero visibility to the naked eye, close to zero light conditions and in any weather.
Helicopters are not the only vehicles that need to detect potential threats, especially in military situations. As Henri Bouma of TNO Security and Safety in the Netherlands pointed out in his presentation, "Historically, warfare was symmetrical. Now, big marine ships have to be able to operate in narrow waters. Small boats can get much closer but the big ships have to react quickly if they get too close. Our goal is to assist the human operators and improve the situational awareness for their images."
One of the first challenges to detecting objects successfully in a harbor environment is to find the horizon -- the boundary between the water and something else -- correctly and therefore define the region of interest in the image. Bouma and colleagues found that region-based methods often fail to do this. Instead, they opted to use edge-based methods.
Horizon patterns are then removed by background estimation. This is also not a straightforward process. "Simple row-based background estimation gives clear artifacts and false detection because the horizon is not usually horizontal in the image," explained Bouma. "Background estimation has to be robust and this cannot be achieved by fitting a simple model like a polynomial. The intensity of the object may also influence background estimation."
The value of the image produced is increased if the harbor environment is tracked over time. Artifacts that are only there for a short amount of time can be removed from the image as they are probably waves.
Bouma and colleagues tested their approach with different object sizes, different cameras at different wavelengths and different environments (for example, near the coast, with clouds and with waves). They found that the sensitivity was improved (greater than 80 percent) and the number of false positives decreased with their approach compared with their previous studies.
Beyond military applications, passive sensing also has an important role to play in spotting and preventing trouble. At the Security + Defence meeting, researchers from Fraunhofer-Institut für Informations- und Datenverarbeitung in Germany reported systems to help identify and track key troublemakers in riots.
"The aim with controlling a riot is to get a de-escalation of violence. Software tools can enhance situation awareness for riot squads," said Uwe Jaager of Fraunhofer. The system that Jaager and colleagues are working on is designed to track individuals in a crowd and identify them. It should help protect entrances and those who guard them.
Juergen Metzler of Fraunhofer added that the system consists of four modules: data acquisition, manual detection, tracking, and situation analysis. The manual detection module can also be changed to an automatic one.
The system has been tested using actors in the roles of rioters and guards. The tests used single-chip color CMOS cameras with a sampling rate of 20Hz, mounted on cranes 25m above the ground.
These tests highlighted many of the challenges in tracking key troublemakers. There are difficulties in poor lighting conditions, and small people are harder to track because they are lost among their taller companions. Similarities between different people pose another challenge, especially in tracking soldiers. One of the most tricky issues to resolve is weak contrast between people and the background.
There are also challenges in ensuring quality of the resulting data set. With many sensors in use there is the risk of information becoming mixed up or dissociated from the time and location at which it was collected. The Fraunhofer team has resolved this issue by synchronizing the data with GPS so that each frame has a GPS-based time stamp. The team has also created benchmark data using the actors in different scenarios with three levels of violence. It has also worked with image processing experts to develop and agree on a set of metrics to evaluate how different imaging systems handle this data.
The work described above was presented at the SPIE Europe Security + Defence symposium in Cardiff, UK, in September 2008. The next meeting will take place in Germany in autumn 2009.
Siân Harris is a UK-based freelance writer.