It has been known for some time that simulation is an essential piece of the test, development and deployment puzzle for connected and self-driving vehicles. With the requirement to drive millions, if not billions, of miles for automated driving system maturity, exposing these systems to high risk and expensive scenarios to ensure safety on real roads and real-world testing alone is untenable for the task at hand.
Simulation enables the required testing miles and scenarios to be tested in a much faster, resourceful manner. It allows developers to fail fast, early on, at a much lower cost and risk, ensuring that crucial system safety and operational maturity can be developed before getting to real roads. It is, however, critical that virtual environments sufficiently replicate their real-world counterparts. This includes the manifold scenarios that need to be tested, which would ensure that any learnings gathered in virtual environments are representative of the results that would be achieved in the real world.
This requires comprehensive simulation capability.
Numerous models and data for vehicles, dynamics, sensors, actuators, traffic, environment, pedestrians, real world actors, scenarios, and much more must be orchestrated across complex simulation capabilities and software. The various components need to meet certain requirements including fidelity, latency and Quality of Service (QoS), and interact seamlessly to ensure a representative simulation of the real-world environment.
The question then is how do we bring together a comprehensive simulation capability, involving complex “inter-operation” between multiple different models, data and software? It is this “inter-operation” that must be mastered.
Simulation capabilities can be made “interoperable” with each other by establishing distributed connectivity between capabilities and employing common formats for models and approaches for simulation across test sites. By doing this, one can produce complex simulations utilising multiple geographically disparate capabilities and navigate around those capabilities with minimal integration effort.
This concept of interoperable simulation was proved as part of Phase 2 of the CAM Testbed UK Interoperable Simulation Project. By making capabilities at Smart Mobility Living Lab: London, WMG and Millbrook Proving Ground interoperable, a real-world self-driving vehicle developer, StreetDrone, harnessed broader and deeper testing insights, and reduced significant resource and integration effort during the testing process.
By employing a DDS middleware-based simulation framework, all partners’ data was protected whilst being able to utilise the distinct simulation strengths across the CAM Testbed UK simulation capabilities.
Phase 3 extends this concept.
Building on the existing framework, this phase will deliver an extended demonstrator, led by KAN Engineering, featuring an industrial level ADAS system from a global customer. The demonstrator will showcase more CAM Testbed UK sites and capabilities and, for the first time, capabilities from outside CAM Testbed UK, from the University of Leeds, will be featured.
In a separate work stream led by Arcadis, supported by Reed Mobility, the project is also exploring the various opportunities for interoperable simulation. This workstream will explore how a future, operational CAM Testbed UK interoperable simulation capability can best serve various customer needs and the CAM sector more broadly.
Interoperable simulation meets the need for comprehensive and flexible simulation capability in the successful development and deployment of CAM. This next phase of work builds on the robust foundation already established, continuing to demonstrate the compelling proposition interoperable simulation brings. This is only the beginning of an emerging new paradigm for simulation, the impact of which will be far reaching and transformative in the at-scale realisation of CAM on our roads.
We look forward to sharing more information with you as the Interoperable Simulation project develops. Sign up to our newsletter to be notified of these updates.