Issue 45 | Uncrewed Systems Technology Aug/Sept 2022 Tidewie USV Tupan | Performance monitoring | Bayonet 350 | UAVs insight | Xponential 2022 | ULPower UL350i and UL350iHPS | Elroy Air Chaparral | Gimbals | Clogworks Dark Matter

69 themselves. Typically, it consists of some combination of perception sensors such as Lidar, radar and EO cameras, the drive-by-wire system, GNSS for localisation, a Cisco wireless link for sending car data back to race control, and a main computer consisting of an Intel CPU and an Nvidia GPU, both supplied by ADLINK Technology, for processing sensor data. “Lidar has been the most critical perception sensor for this competition, given the speeds the cars are moving at and the importance of detecting turns or other cars if they’re racing head to head as they did for the CES 2022 challenge,” Mitchell noted. “Luminar is the official Lidar supplier to the IAC. They have a great product, and the teams have been pleased with it, but technically speaking there’s a wide range of sensor types they could choose from.” Professor Sergio Matteo Savaresi, team leader of Team PoliMove, said, “To win the head-to-head challenge at CES, we needed a full autonomy software stack for optimising localisation, control of dynamics, perception and many other elements and functions. “What we also needed in the land speed record attempt was in some ways simpler but in others more complex and critical. The two things most important in autonomously handling a really high- speed straight are maintaining control, which is not easy at 300 kph, and prediction of trajectory. “GPS can give you a good position output, but when you’re moving close to 100 m/s, a seasoned human racer can automatically predict well what they’ll need to do in 100 m but the computer can’t without the right programming. “If it makes errors in that, the controls pull in the wrong direction and there are huge risks of instability in the car. That’s the technical challenge we had to face, and no-one’s ever deployed an autonomous car in that way before.” The ECU was left largely untouched for the record drive, other than ensuring it kept the Honda engine producing 500 bhp. The perception software meanwhile was configured to enable sensor data to be processed by the main computer and output control decisions to the steering actuator in a continuous loop updated every millisecond. “We are using an ROS-2 framework based on a Linux machine,” Savaresi added. “Within that, we use a high-level Simulink software code as well as more traditional C# code. “ROS-2 is flexible like that. You can use many different programming tools, which is important because you have many different issues to solve in high-speed autonomous driving. “Processing Lidar signals for instance is very intensive from a data point of view, so you need a very capable GPU as the platform for it. Also, doing traditional things alongside that such as managing IMU data requires a lot of different software skill sets.” The largest static display in the expo hall was Aergility’s ATLIS VTOL UAV, which the company unveiled in its full-size form, ready for long-range heavy-lift cargo delivery in disaster aid relief, defence and other critical supply applications. “The ATLIS’ VTOL capability means it doesn’t need landing ports – it directly loads and unloads supplies exactly where they need to be,” said Jim Vander Mey. “That includes medical and food supplies to remote islands, hundreds of thousands of villages in Africa without easily serviceable roads, and even equipment and water to frontline military operating bases, without putting human pilots, truck drivers or logistics officers at risk and without incurring high costs.” The ATLIS lifts off using battery energy to power six fixed-pitch rotors, then uses a turboprop to transition into forward flight, with the e-motors transitioning into autorotation. In this mode, the rotors still provide lift but do so passively, without consuming any net power. “That’s our patented ‘managed autorotation’ technology. When we bank right, we draw power from the right-hand rotors, and put power into the rotors on the left,” Vander Mey explained. “So we use no control actuators for banking, roll or pitch, and in cruise mode the AUVSI Xponential 2022 | Show report Uncrewed Systems Technology | August/September 2022 The Aergility ATLIS long-range heavy-lift UAV