Unmanned Systems Technology 009 | Ocean Aero Submaran S10 | Simulation and testing | Farnborough report | 3W-110xi b2 TS HFE FI | USVs | Data storage | Eurosatory/UGS 2016 report

32 A s unmanned systems become more complex, with a growing number and range of onboard sensors, faster data rates and even machine learning, so the process of developing and testing them is becoming more challenging. Also, the regulatory environment is requiring developers to demonstrate that their systems are safe, both for users and bystanders, so this process is becoming an integral part of their manufacture. Simulation is one area that has grown in importance in the development of unmanned systems, and this can be used at various stages, from testing out algorithms for control systems or sensor fusion to running software on a simulated model of a microprocessor. It is also replacing physical testing in wind tunnels (see sidebar, page 39). The increased use of simulation is being driven by improvements in the hardware and software. On the hardware side, the latest server processors have up to 44 cores on a chip, and many multi-chip servers can be combined in a server farm or cluster. Access to supercomputers, which use the latest graphics processor units (GPUs) to handle the calculations and render the results, is also becoming more cost- effective than before. The capability of the software has been enhanced to take advantage of parallel elements in the problems being simulated. If significant parallelisation of a problem is possible, it allows the As simulating and testing autonomous vehicles grow in importance, so the tools are becoming more advanced and cost-effective. Nick Flaherty reports Pilotless studies Simulating the design of an autonomous vehicle with high-level models and hardware in the loop shortens the test time and increases the reliability of the end system (Courtesy of D-Space) August/September 2016 | Unmanned Systems Technology