Unmanned Systems Technology 011 | C-Astral Bramor ppX | IMUs | Autonomous farming | UAV Turbines UTP50R | London Show report | Advanced materials | Las Vegas Expo report

7 Platform one Unmanned Systems Technology | December/January 2017 UUV modelling Study to simplify control Autonomous control of unmanned underwater vehicles is a challenge because they are underactuated systems with non-linear dynamics, and are subject to uncertainties in hydrodynamics, currents and obstacles (writes Peter Donaldson). Model predictive control (MPC), which relies on a computer model of the vehicle and its response to control inputs, is the focus of much research. However, creating such models and implementing control schemes based on them is complex and computationally intensive, so researchers at the Veermata Jijabai Technological Institute in Mumbai, India, are developing a simpler approach. They are working on a non-linear, non-holonomic and simplified model of an AUV with six degrees of freedom: pitch, roll and yaw plus heave, surge and sway. The model includes a rudder and an elevator, and is composed of two reduced-order models, one to control the vehicle in the vertical plane and the other in the horizontal. If all a vehicle’s degrees of freedom are controllable, it is said to be holonomic, but with just two control surfaces the VJTI’s model is non- holonomic. This simplified model helps to tackle issues that arise in conventional MPC for non-linear systems with more than two or three dimensions or degrees of freedom, such as reduced controllability and increased computational effort, according to the researchers. The new approach uses a time- dependent linearised model formulated at every time instance, allowing the controller to manage non-linear dynamics effectively while allowing for constraints on control inputs. Simulated in MATLAB, it proved able to track the desired height and yaw references without affecting remaining states such as roll, say the researchers. Its use of time-varying dynamics to solve optimisation problems enabled it to perform better than a proportional integral derivative controller, which is linear in nature. Future work will focus on integrating hydraulic actuator dynamics and estimator design for an MPC scheme better suited to controlling a more realistic AUV model. Component maker NXP has ambitious plans to replace Lidar laser ranging systems in driverless cars (writes Nick Flaherty). It sees an array of eight single- chip radar systems as being enough to provide a ‘cocoon’ around a vehicle to provide all the required data. That would be inherently less expensive than several Lidar sensors. The move comes as NXP integrates with Freescale Semiconductor, which supplied many of the automotive controller chips. “The next generation- one chip design is underway,” said Lar Regen, NXP’s chief technology officer for automotive. “NXP has done this for other markets such as car radio – we had the tuners and baseband for radio, and our way forward was to get all of that onto CMOS for a single chip [called Mercury]. “We have done the same for car- to-car communications at 5 GHz, combining the RF and baseband into one chip. Now we are doing that for 77 GHz. We are at a two-chip solution at the moment.” That will lead to what he calls a “perfect radar cocoon” based on 77 GHz radar rather than 24 GHz systems. “Some car makers are looking at 20 radar systems around the car to make the Lidar systems redundant,” he said. The timescales for the radar are not yet public, but with designs underway the prototype parts should be available in 2018, with qualified production parts by 2020, which fits with the plans of large car makers for mainstream production of driverless cars. Single-chip option to Lidar Driverless car sensors NXP’s roadmap for a single-chip radar system