Unmanned Systems Technology 033 l SubSeaSail Gen6 USSV l Servo actuators focus l UAVs insight l Farnborough 2020 update l Transforma XDBOT l Strange Development REVolution l Radio telemetry focus

7 Platform one A group developing an open source stack for autonomous platforms is adopting the latest tools based around the Data Distribution Service (DDS) standard (writes Nick Flaherty). DDS uses a technique called publish- subscribe to connect different hardware and software blocks without the traditional primary-secondary bus technologies. The DDS implementation from Real Time Innovations (RTI) is being added to the stack development of the Autoware Foundation. The DDS technology will be used with autonomous vehicles as well as simulation projects. Autoware was set up by machine learning specialist Apex.ai, board maker Linaro and systems integrator Tier IV. The main focus for DDS will be Autoware.ai, the autonomous vehicle project, but RTI will also use its experience with ROS2, the open-source robotics project, for the other two projects in the foundation. The first of those is Autoware.IO, an interface project that can be extended with proprietary software and third-party libraries for device drivers for sensors such as Lidar and radar. The second is Autoware.Auto, a rewrite of the original Autoware but now based on ROS2. Autoware.ai will use the Connext 6 DDS software. This adds a data representation that significantly reduces latency and CPU usage for large and small data samples, which is critical for sensor fusion applications that need to minimise the time between when an event happens and when it is recognised by the sensor fusion, localisation and path planning algorithms. Any delay reduces the fidelity of the real- world model, and makes an autonomous vehicle’s understanding of what is happening around it less accurate. This is a key part of the Lidar, radar and image data processing subsystems. Connext 6 also uses a shared memory transport to nearly eliminate end-to-end latency. This allows developers using Autoware.Auto to use the same interface to communicate between applications running on the same CPU and across networks, while still maintaining the performance they expect and need from software modules running on the same platform. This reduces development costs by avoiding the use of another protocol, while DDS enables IP portability, using software on a common CPU and distributing it to other processors in a second architecture without having to redevelop the software. DDS also includes security that can be safety-certified – a key requirement for the stack being developed for Autoware.ai. RTI is working on 250 autonomous vehicle projects, and works with more than 50 commercial autonomous system developers including Aptiv, Baidu Apollo and Xpeng Motors.   DDS tools picked for project Autonomous vehicles Unmanned Systems Technology | August/September 2020 The DDS is being adopted by the Autoware open source stack projects for autonomous vehicles