Issue 41 Unmanned Systems Technology December/January 2022 PteroDynamics X-P4 l Sense & avoid l 4Front Robotics Cricket l Autonomous transport l NWFC-1500 fuel cell l DroneX report l OceanScout I Composites I DSEI 2021 report

42 Focus | Sense & avoid systems aircraft they advise. This is accomplished through an automated network of data services, to relay flight plans and airspace constraints between UAVs and their operators. Tests are now showing how ‘explainable’ artificial intelligence (XAI) can improve trust in those human air traffic controllers in operating automation- enhanced ATM systems. Explainable AI provides more visibility of the decision-making process to give more confidence in the results. These AI applications are often highly complex, with deeply recursive statistical methods forming the core of the learning process, and it can be difficult for people to understand why the AI-enabled application proposes a certain solution and how it reached a particular decision. This is referred to as a ‘black box’, because the inputs and operations of the AI application are invisible to the operator.  In a project called TAPAS, XAI applications have been developed that use a technique called deep- queue learning. These applications are helping the human operators with demand capacity management for the European ATM network by using visual analytics tools and a customised flow management position to identify the different craft in the air. The XAI applications and the flow have been tested using a network management simulation tool platform to execute air traffic flow management scenarios at varying levels of automation. These work with a publish-subscribe data network to create a highly realistic environment to help determine how the explainability and transparency information could help the operators understand how the automated solution were being produced, and why. Feedback and the results from the project have helped to confirm that explainability is a key element in developing trust in AI applications, and the XAI algorithms are being adapted to look at how they can be used for managing craft separation in the air. UTM systems Machine learning algorithms are helping UTM systems to manage UAVs from multiple operators, The USS (UAS service system) from different UAV operators can be managed by the central UTM system to prevent individual aircraft from approaching each other and reroute them in real time, for example if there is an accident and emergency services need access for their UAVs or an air ambulance. Every operator has their own USS and management system, so they know December/January 2022 | Unmanned Systems Technology A central management system can provide sense & avoid capabilities for UAVs using position and velocity data sent over wireless links (Courtesy of Airwayz) ACAS Xu/EUDAAS TCAS II- equipped intruder Cooperative surveillance DAA- equipped intruder Transponder/ADS-B only equipped intruder No transponder/ no ADS-B OUT Non-cooperative surveillance (radar, electro- optical surveillance or other sensors) Maneuverability The SESAR project, part of EUDAAS, is looking to detect a wide range of airborne systems (Courtesy of SESAR)