UBER: An Unreal Engine Based Simulation Platform with Extensibility and Real-time Capability*

In this paper, we propose a real-time platform called UBER, which stands for Unreal Engine Based simulation platform with Extensibility and Real-time capability. It provides a visualization way to the online tests of unmanned vehicle simulation with Unreal Engine 5 (UE5). By building a TCP communication module to exchange message between two ends in UBER, we successfully solve the problem that no code script can be operated in blueprints of UE5. Excellent physics components in UE5 makes simulation visualization more real and beautiful compared with other Reinforcement Learning environments, and by accessing the python scripts outside of blueprints, various of elements, such as scenario components and environmental settings, etc., can be easily modified. Furthermore, a series of reinforcement learning and formation control algorithms are implemented on UBER to show its extensibility and real-time capability, proving that our TCP communication method brings forth a good solution to the problem that UE5 does not support python scripts within blueprint.