Issue 39 Unmanned Systems Technology August/September 2021 Maritime Robotics Mariner l Simulation tools focus l MRS MR-10 and MR-20 l UAVs insight l HFE International GenPod l Exotec Skypod l Autopilots focus l Aquaai Mazu

40 Focus | Simulation tools automated driving functions and the various driving simulation frameworks available, ASAM is proposing an open simulation interface (OSI). This will contain an object-based environment description using the message format of the protocol buffers library developed and maintained by Google, and consists of two individual top- level messages defining the GroundTruth interface and the SensorData interface. The GroundTruth interface gives an exact view on the simulated objects in a global coordinate system. The messages are populated using the data available internally and then published to external subscribers by a plug-in running in the driving simulation framework. The SensorData interface describes the objects in the reference frame of a sensor for environmental perception. It is generated from a GroundTruth message and can either be used to directly connect to an automated driving function using ideal simulated data, or serve as an input for a sensor model simulating limited perception as a replication of real- world sensor behaviour. The aim is to be able to connect any automated driving function to any driving simulator with ease. This will simplify the integration of the models and boost the use of virtual testing. This is vital for the supply chain, as a Tier 1 supplier and an OEM might use different simulation tool chains, which are different again from the tool used by the sensor developer. ASAM is now working on four packages to develop the OSI with physical sensor modelling with a better GroundTruth, traffic participants and roadside infrastructure, improving the real-time performance of the interface and harmonising with other software standards. Fidelity There are multiple requirements for fidelity for simulation engines. The sensor models need to accurately represent the hardware, and the simulation has to accurately model the environment. Real-world data needs to be fed back into the simulation to refine the accuracy and ensure that the model accurately represents the real world. The fidelity between different levels of abstraction also has to be demonstrated, from the cycle-accurate back-up to high-level abstraction for faster simulation. Simulation in the real-time control loop Simulation technology is even being integrated into the real-time control loop to make predictions. Researchers in Germany for example have developed a software module that forecasts what happens around a driverless car.  The module calculates all the theoretically possible movements for each road user for 3-6 seconds in the future. For example, as a car approaches an intersection and another vehicle emerges from another road, it is not yet possible to tell whether it is turning right or left. At the same time, a pedestrian runs into the road directly in front of the car, while a cyclist is standing on the other side of the road. Based on these scenarios, the system determines various movement options for the vehicles, along with possible emergency manoeuvres with braking and acceleration to provide a safe result without injuring pedestrians or other drivers. Only if a trajectory is identified without a predicted collision and an emergency manoeuvre being possible at the same time would it be used by the autonomous vehicle. This is made possible by using simplified dynamic models that August/September 2021 | Unmanned Systems Technology Modelling a FLIR thermal camera (Courtesy of Ansys) The aim is to be able to connect any automated driving function to any simulator with ease, as this will boost the use of virtual testing