Unmanned Systems Technology 007 | UMEX 2016 report | Navya ARMA | Launch & recovery systems | AIE 225CS | AUVs | Electric motors | Lethal autonomous weapons

27 clearance of 20 cm, and weighs between 2100 and 2300 kg depending on the equipment in the vehicle. It uses a 33 kWh battery pack that provides a top speed of 45 kph and a cruising speed of 25 kph. It travels both backwards and forwards, and both ends of the vehicle are identical. This has had implications for the design, as in many places around the world road vehicles are certified for operation with a maximum of two sets of two rear red lights. This has meant that the ARMA system has three lights on each side that can either be three white high-brightness LEDs or two red LEDs for braking. The lights of course are not required for operation, but to warn other road users about the presence of the vehicle and for the certification authorities. Navya’s focus on the mass-transit market has influenced the overall engineering design of the ARMA platform, which is based on a steel chassis with steel tubing to provide a structure, a fibreglass cover and thermoplastic details for branding by operators such as bus companies. But the design is also flexible to allow different versions to be built to carry more or fewer passengers. This will allow Navya to quickly produce a two-person ‘pod’ similar to that being developed by Google, if requested, although the company insists that it is not looking to compete with traditional car makers. Sensors Navya believes the key to a Level 5 autonomous ground vehicle rests in the architecture used for the sensors. “We are convinced that if we are to succeed then we need to multiply the number of sensors – no specific sensor is able to give all the things we want,” says Navya’s CEO Christophe Sapet. The company insists on having three sensors providing data, for example from two Lidar sensors and a camera, with each sensor allocated a level of confidence in the data. This is a complex combination of the relation of a data point to the previous one and over time. The further the data falls outside the expected range, the lower the confidence. This then provides an overall level of confidence in the data from all the sources without having to use techniques such as redundancy and voting by three separate data sources. The data is all time-stamped with a resolution of 1 ms, combined with the position data from the GNSS navigation system and odometer, and fed into the vehicle’s central controller to determine its speed and direction. If any of the sensors provides a low level of confidence (or fails) the redundancy means the other sensors allow operation to continue. If the confidence level is too low on all the sensors then the vehicle will come to a safe stop. Lidar Navya has found Lidar to be the most effective sensor, as it provides a high- quality signal for obstacle detection and to build a map of the area around the vehicle. The two Velodyne VLP16 16-channel Lidar ‘puck’ sensors on the ARMA use a 903 nm laser that scans in a circle, taking 300,000 measurements a second. These are mounted on the roof, angled downwards (but also measuring upwards) with a 180° field of view in the direction of travel and behind. They have a range of up to 100 m, a power consumption of 8 W and weigh 830 g. They do not have visible rotating parts, allowing them to be rated to IP67 for use in the wet. One sensor monitors the path Navya ARMA | Dossier Unmanned Systems Technology | April/May 2016 The top Lidar sensors from Velodyne and cameras provide data for mapping We are convinced that if we are to succeed then we need to multiply the number of sensors – no one sensor is able to give us all the things we want