UST030

38 artificial neural network, so that machine learning can train the car’s computer to identify when something is about to set foot on the road. Systems of multiple thermal cameras per autonomous vehicle are therefore being tested as anti-collision systems, enabling the onboard computer to brake or slow down to avoid a collision – in line with calls for better self-driving safety systems. In addition to improving over the key weaknesses of EO sensors and the range limitations of radar, Lidars tend to collect a somewhat sparse point cloud in a vertically limited space. That is problematic for object classification, but thermal cameras have proven so far to always collect enough pixels in order to know what an object is. Also, Lidar pulses must reflect off of an item in the scene and return to the sensor. Fog can make that very troublesome, whereas thermal cameras need only a single path from the scene to the sensor. Optical gas imaging Over the past few years, the use of LWIR and MWIR cameras for detecting gas leaks in hydrocarbon pipelines and facilities has grown rapidly, thanks to the development of smaller, lighter optical gas imaging (OGI) systems designed to be integrated into SWaP-constrained platforms such as UAVs. Many gases are invisible to the naked eye but visible in the IR band. For example, a MWIR sensor with an InSb focal plane can detect methane – and gases including ammonia, benzene and sulphur dioxide – between 3 and 5 μm. By contrast, sulphur hexafluoride (a potent greenhouse gas which can be toxic to humans) is only visible at 8-12 μm, so a LWIR camera is needed, typically with a quantum well infrared photodetector made from gallium arsenide. Rather than the wide FoV thermal imagers used by UAVs for agriculture, conservation or firefighting, a typical UAS OGI camera has a 23-25 mm lens with a 24 x 18 º FoV (or a 38-40 mm lens perhaps with a 14 x 10 º FoV ), to enable inspections of pipelines, gas fields and other similar infrastructures over short ranges. Development and integration of HOT detectors has also helped cut the cost and weight of cooling circuits in OGI sensors. This is markedly different from the IR cameras being supplied to UAVs in defence and security. Many of these – both long-wave and mid-wave – now integrate continuous zoom lenses, a rarity for thermal cameras. They are used on heavier, fuel-powered aircraft that need the ability to survey large regions as well as closely inspect single targets. In some cases this results in thermal cameras that weigh upwards of 8 kg and with focal lengths up to 900 mm. However, the lightness and low prices of OGI cameras for UAVs has enabled wide commercial adoption of such systems, as they can be integrated into small, low-cost multi-rotor UASs. This is allowing more and more industrial companies to locate leaks of potentially toxic or explosive gases without sending workers into dangerous environments. February/March 2020 | Unmanned Systems Technology Tests indicate that thermal cameras improve significantly over EO cameras, and sometimes over Lidar and radar, for example when experimenting amid excess glare (Courtesy of FLIR) Many harmful gases are only visible at certain points of the infrared spectrum, making optical gas sensors an increasingly popular payload choice for UAV operators in the oil and gas industry (Courtesy of Sierra Olympic)