90 Focus | IMUs, gyros and accelerometers surface for high-fidelity transmission of light from the source and via a waveguide into the fibre. As the FoC optical chip typically integrates what would otherwise be 15 to 20 loose components (albeit the coil, laser and detector sit outside of that), much of the production complexity is outsourced to traditional chip-fabrication processes. The mass production of those chips also makes FoCs significantly less pricey than equivalent traditional FOG designs. Future work in FoCs is expected to conceive new designs that incorporate the coil, laser and detector inside the chip – creating, in essence, a microelectromechanical FOG. External aiding As indicated, advances in external aiding integrations and sensor fusions are occurring across the uncrewed world at a comparable pace to improvements in MEMS and FOG IMUs themselves. UGV applications are arguably the fastest growing area of uncrewed systems currently, and as UGVs generally operate with significantly reduced visibility of the sky, compared with UAVs, navigation and mapping systems that go beyond traditional GNSS-IMU integrations are imperative (even two-storey buildings are enough to block some small delivery robots from receiving GNSS signals). Resorting to higher-end IMUs can be an option if the increased cost and SWaP are no bar, but for many INS manufacturers, navigation software engines (including sensor fusion models and error models) have been developed for decades now to function with MEMS devices, not with FOG IMUs. The most common approach has been to optimise programming, error modelling, calibration and sometimes MEMS IMU designs to squeeze more performance out of MEMS. Some companies will alternately mount multiple triple-axis MEMS chips (such are their small size nowadays) on a board to average out the data, and come up with higher assurance and precision of inertial readings. Even more common is the consistency of solutions leveraging non-GNSS aiding sources (as complementary or replacements to the satellite signals), which have been increasingly available over the past few years. Such sources include ultra-wideband (UWB) systems for indoor positioning, with one trial using UWB sensors throughout a multi-storey car park to enable autonomous valet-parking robotics, as well as Lidar integrated into 3D SLAM arrangements besides Lidar-inertial odometry configurations. Similarly, cameras are being leveraged and fused with inertial data to output forms of visual odometry across UGVs and UAVs, including situations where a camera was already present on the vehicle, eliminating the need to install significant new hardware onboard. Integrating IMUs with these new forms of aiding sensors, with their notably different parameters, error models and applicability (depending on the environment) to GNSS can be challenging. One successful solution is to use a genericising framework as middleware that can conditionally ‘subscribe’ the INS’s processor to data from external aiding sensors, so long as the data format is acceptable and accuracy parameters are also given to enable appropriate weighting in the Kalman filter, along with timestamps and, if applicable, a lever arm (a technical term for the distance between the actual sensing points of the sensors). Such a framework can enable a wide variety of sensors to be easily read by an INS and incorporated into an autonomous vehicle’s localisation calculations. Hence, this approach is expected to experience significant uptake in the future. The future As demand for robust navigation, mapping and geo-pointing rises among professional uncrewed vehicle users across government, business and defence, MEMS manufacturers are expected to double-down on the performance capabilities of their solutions, with a major target being the creation and commercialisation of the first navigationgrade MEMS IMU. Such a device would also be capable of gyrocompassing – another milestone for MEMS, which would put them on a par with FOGs. FOG developers are working on a wider spectrum of devices, which may blur what were once quite clear lines between MEMS, FOG and RLG applications. This, combined with the variety of INS configurations and external aiding sources growing across the uncrewed world, is likely to spur considerable growth in software algorithms (including open-access solutions) for fusing inertial data and Kalman filters with myriad combinations of other sensors – depending on which combinations make the most sense in the sky, on the road, in water, underground, and everywhere else that uncrewed systems can go. Acknowledgements The author would like to thank Walt Johnson and Thomas Bennett of Inertial Sense, Arthur Tua and Olga Pogorelova of Fizoptika, Ian Moore of MicroStrain by Hottinger Bruel & Kjaer, Eric Whitley of Silicon Sensing Systems, David Cunningham and Sébastien Ferrand of Exail, Iain Clarke of OxTS, Xavier Orr of Advanced Navigation and Richard Ryu of Fiberpro for their help in researching this article. February/March 2024 | Uncrewed Systems Technology Future IMUs will include MEMS that weigh less than a gram, while packing in increased performance through rigorous software improvements (Image courtesy of Inertial Sense)
RkJQdWJsaXNoZXIy MjI2Mzk4