基于UE的水下机器人动态仿真平台设计与实现

Design and Implementation of a Dynamic Simulation Platform for Underwater Vehicle Based on Unreal Engine

摘要:

仿生水下机器人需要大量的训练数据来模仿水下生物的各种任务；然而，真实物理环境中的高成本使得仿真成为应对这一挑战的可行工具；目前，多数的仿真平台对于仿真环境的设计较为简单，缺乏在保证模拟真实性的条件下构建大型复杂场景的能力；为了解决仿真平台环境建模能力的不足，提高仿真效果的可视化程度，采用虚幻引擎（Unreal Engine，UE）、AirSim和分布式算法仿真平台（Distributed Algorithm Simulation Platform, DASP）相结合的方案，设计了一款应用于仿生水下机器人的动态交互仿真测试平台；并通过LQR控制器控制仿生机器鱼跟踪目标轨迹对仿真平台的可视化和有效性进行了测试；仿真结果表明该仿真平台在交互性、真实性、可视化和仿真效果等方面表现出较高的优势。

Abstract:

Bionic underwater vehicles require large amounts of training data to mimic the various tasks of aquatic organisms; however, the high cost in actual physical environments makes simulation a viable tool to address this challenge; currently, most of the simulation platforms are simple in the design of the simulation environment and lack the ability to build large and complex scenarios under the condition of ensuring the realism of the simulation; in order to solve the shortage of simulation platform environment modeling ability and improve the visualization of simulation effect, a dynamic interactive simulation test platform applied to bionic underwater vehicles is designed by combining Unreal Engine (UE), AirSim, and distributed algorithm simulation platform (DASP); and the simulation platform was tested by controlling the bionic robot fish tracking target trajectory through the LQR controller visualization effectiveness of the simulation platform was tested by controlling the target trajectory of the bionic robot fish with the LQR controller; the simulation results showed that the simulation platform showed high advantages in interactivity, realism, visualization, and simulation effect.