UST030

36 Focus | Thermal imaging sensors at roughly 77 K to mitigate this self- generated noise, entailing considerable power and weight to keep it cool. Having a cooled core is critical to reducing the thermal ‘noise’ – the ambient heat that can throw off the piezoelectric measurements – around the detector. The best way to minimise that is to lower the temperature inside the core to below the point at which the radiation can be detected. If a thermal camera core were operated at room temperature, for example, there would be constant, random thermal activity smothering the actual IR radiation signals coming from objects being viewed through the sensor’s optics. Heat around the detector must therefore be removed until it no longer interferes with the band of the IR spectrum in which the sensor is operating. As mentioned though, cooling systems take up weight, space and power, which is a drawback as regards integrating them into UAS platforms. Fortunately, uncooled LWIR sensors are becoming increasingly available, thanks to innovations such as more sensitive temperature sensors for detecting thermal noise, and improvements in how the detector array is designed to reduce thermal ‘crosstalk’ between resistors. However, the development of uncooled camera cores for the MWIR 1-3 μm range has proved challenging for commercial development, largely because of the difficulty in detecting and extracting interferences in such a narrow wavelength band. It also means a smaller pixel pitch (the distance between resistors on the focal plane array) for MWIR detectors compared with LWIR detectors, so more cooling is needed to combat the increased thermal conductivity across the former. In terms of new detector materials, type-2 strained layer superlattices (SLSs) represent perhaps the most promising innovation, as they bring a number of advantages when used in the MWIR and LWIR bands. For example, they have been found to exhibit much greater uniformity and stability across temperatures than many other materials in the LWIR band. They are also capable of very short exposure times compared with other detectors. Tests indicate that they can take images in 0.16 ms between -20 C and 150 C, whereas InSb varies from 0.24 to 2 ms across this temperature range. This means SLSs can reduce blurring – and therefore increase accuracy – in thermal imaging by significant margins. Automatic braking systems Research is currently being carried out on using thermal video to develop an enhanced form of ABS for self- driving vehicles. Interest in new ABS architectures has grown considerably since Uber Corporation’s accident in March 2018, in which the test vehicle used Lidar, radar and cameras for perception but only recognised a cyclist as a human 1.3 s before the collision. In essence, thermal imaging greatly improves both the detection ranges and prediction capabilities of road vehicles, particularly at night or in scenes with what’s called a high dynamic range, such as when the vehicle is driving towards the setting sun or oncoming headlights. Computer vision using EO cameras is greatly hampered in such situations, with sections of the image being blotted out by brightness or darkness. This approach to obstacle detection and avoidance can therefore completely miss humans or animals walking out into a road, raising serious questions about current perception architectures for self-driving cars, which rely largely on little more than vision and radar. LWIR cameras, on the other hand, by virtue of only picking up heat in the 7-14 μm band and ignoring visible radiation in the 0.4-0.7 μm range, can detect biological heat signatures regardless of fog, or bright or dark conditions. Even among manned vehicles, accidents involving pedestrians tend to occur at night or in poorly-lit areas, with speed limits of no more than 45 mph. As a result, LWIRs are proving highly effective at enhancing the reaction times of ABS in highly dynamic scenes. Initial research has examined the use of night- vision cameras to feed the thermal video in real time to the driver’s dash, and sound an alert when a person or animal is detected entering the road further ahead than using the visible-light spectrum. However, that interface has been judged as an unsafe distraction for drivers, so it has been reconfigured to feed the thermal information to an February/March 2020 | Unmanned Systems Technology Different detectors can reveal details from different infrared bands, enabling critical survey operations such as monitoring fire perimeters and hotspots (Courtesy of Overwatch Imaging)