Unmanned Systems Technology 036

36 Focus | IMUs, gyros and accelerometers secure, fast and intuitive control loops to allow comms between flight controllers and payload gimbals. Gimbals must now be able to lock on intelligently to targets, be they moving ones such as suspicious vehicles or vulnerable groups of people, or stationary ones such as landmarks to be 3D-modelled. Without a secure and comprehensive interface on the IMU to link through to the gimbal’s motor controllers, the latter will not be able to exploit the former well enough to justify the investment. Furthermore, customers in the UAV gimbal-tracking space have long since surpassed the 100-250 Hz update speeds typically used in inertial navigation systems. This has been spurred greatly by uptake of the latest processor chips in their autopilots and motor controllers. As a result, IMUs for these applications increasingly need to be capable of more than 1000 Hz (450- 650 Hz having been the peak of speed requests only two years ago), with some now reaching 8000-10,000 Hz. Continuous improvements in hardware and software are thus needed – and indeed are being researched – across MEMS IMU developers to ensure the necessary response rates in their accelerometers and gyroscopes for agile gimbal locking in dynamic UAV flight. As to how continuous improvements in IMU software algorithms are actually achieved, much of the methodology remains concealed as IP by development houses and their labs. However, it is noticeable that the companies most pushing the performance and SWaP optimisation of MEMS IMUs are constantly engaged in exhaustive test regimes. Testing to defence industry standards such as Mil-Std-810G for shock and Mil- Std-461E for EMI has become common for products intended for the autonomous vehicles, with manufacturers at the higher end going above and beyond the minimum requirements of these standards. For instance, many IMU manufacturers have set up their own testing facilities to rapidly trial new hardware configurations or software patches for their MEMS products and carry out highly accelerated lifetime testing or stress screening. The aim is to predict how a system will perform (and be improved to better perform) over years of use. To extensively characterise bias stability and other critical performance parameters, some companies now perform hundreds of hours of testing over multiple temperature and vibration cycles. Also, by repeatedly running in-house tests, companies can amass data to use in simulating components and changes for future products, and experiment with new combinations of tests to gauge the effects of modifications in different ways, enabling rapid customisation towards different requirements from end-users. More high-end MEMS IMU developers are also investing in their manufacturing processes to ensure bulk assembly of their sensing elements, processors, boards and housings. Newer automation systems enable greater consistency between batches and reduced per-unit costs, as well as incorporating helpful production processes or configurations. For example, assembly time February/March 2021 | Unmanned Systems Technology Running an Allan Variance test can be critical to gauging an IMU’s random drift error from noise and hence its reliability over time (Courtesy of Gladiator Technologies) To characterise bias stability and other parameters, some companies do hundreds of hours of tests over multiple vibration cycles