UST 031

49 UGVs | Insight plugging in and fastening payloads in different locations and positions. With a kerb weight of 3.08 t, it can carry up to 3175 kg of payload. To provide the necessary power for different surveillance and ordnance payloads, it runs on a diesel-electric series-hybrid powertrain. The platform can offload power at 28 V DC (a low- voltage option providing up to 6 kW) or at 320 V DC for up to 30 kW. With this battery power and a tracked suspension system, the RCV-L can move at up to 88.5 kph, with a typical operating speed limit of 48.3 kph. It can operate over a range of terrains. Warehouse organisation In our previous issue ( UST 30, Feb/Mar 2020) we featured the RISE heavy-lift logistics UAV that was developed and tested in Sweden at the start of this year as part of a larger warehouse logistics service. As mentioned in the article, that logistics service also included the use of autonomous forklift UGVs for lifting, delivering and stacking warehouse stock, which are being developed by researchers from the University of Trento in northern Italy. “We started with a commercial AGV from Jungheinrich, then added an electromechanical and remotely operated forklift system,” says Stefano Divan, a PhD researcher at the university. “We then fitted electric motors on the wheels for independent rotation, and perception sensors for SLAM navigation and to detect and avoid people and other objects.” Professor Luigi Palopoli, who led the vehicles’ development, adds, “We deliberately avoided any excess complexity in mechanical design or perception sensors, because these UGVs have to work autonomously in an industrial setting. Most of the key enabling work lay in the development of the intelligent software. “This included such tasks as programming the navigation systems to manage difficult traffic conditions on the warehouse floor, collectively optimise travel routes and perform other problematic tasks without needing supervisory intervention.” Autonomy is key for improving warehouse logistics and moving beyond the guidance-tape and radar-blocking mirrors needed for standard AGVs, or constant micromanagement through fleet management consoles, hence the development of SLAM navigation. Similarly, it was critical to ensure the UGVs could avoid moving obstacles such as workers. This avoidance capability was ensured by embedding into the vehicles’ autopilots knowledge of how people move about. For each UGV to detect objects and localise itself in its environment, three Lidars from Slamtec are mounted on the system. One is installed atop the 2 m forklift rail to scan without excessive interference from ground obstacles, while the other two are integrated at the bottom of the vehicle – one at the front and the other at the rear – to measure the distance and size of ground obstacles that pose collision risks. Also, an Intel RealSense depth camera installed on the front helps by providing richer information for object classification, which Lidar can fall short on when used by itself (in terms of colours, textures and so on). “The range of our depth camera only really runs to about 3 or 4 m, but the forklift UGV is designed to move at a slow speed, so we don’t need to see very far in front. For example, a 20-30 m range would be far in excess of what it needs,” Divan says. The traction and steering rests on two brushless DC motors from Crouzet, which are powered by a 48 V battery pack, and each outputs 600 W in normal operation. Each also has a gearbox to reduce the speed of the motors while preserving torque, as the 300 kg UGV must be able to transport up to 1 t and cannot travel faster than 1 m/s. And while the UGV’s centres of gravity and balance are consistent regardless of loads being lifted at any operating speed, the researchers also conducted significant algorithmic studies and iterations to minimise the degree of ‘jerk’ from the motor, to avoid generating stray momentum that could pose safety hazards. This configuration was chosen to match the specific project requirements for speed and payload capacity, but Prof Palopoli adds, “The planning and control algorithms could be rapidly mapped onto any similar kind of UGV. They’re written with modular structures so that mission- specific information and constraints can be replaced depending on the unmanned robotic vehicle and the application.” Multi-role urban mobility While autonomous systems for public transport and last-mile delivery are widely in development, very few are aimed at tackling both at once, given the design differences they typically require from their vehicles. However, in the interest of improving the transport of people and goods around cities at the same time, Switzerland- based Rinspeed has unveiled its concept for a driverless road vehicle with an interchangeable internal chassis. The Rinspeed MetroSnap has a patent- pending system of internal ‘pods’ Unmanned Systems Technology | April/May 2020 A modified AGV forms the basis of this autonomous forklift robot, developed by researchers at the University of Trento (Courtesy of the University of Trento)