GPU based generation of state transition models using simulations for unmanned surface vehicle trajectory planning
GPU based generation of state transition models using simulations for unmanned surface vehicle trajectory planning
复制标题
DOI：
10.1016/j.robot.2012.07.009
复制
发表时间：
2012-12
期刊：
影响因子：
--
通讯作者：
Atul Thakur;P. Svec;Satyandra K. Gupta
中科院分区：
文献类型：
--
作者：
Atul Thakur;P. Svec;Satyandra K. Gupta
本文介绍了基于GPU的算法来计算无人水面航行器（USVs）的状态转换模型，使用6自由度（DOF）的车辆-波浪相互作用的动力学模拟。状态转移模型是马尔可夫决策过程（MDP）的一个关键组成部分，MDP是一个自然的框架来制定运动不确定性下的轨迹规划问题。USV轨迹规划问题的特征是由于海浪而存在较大且有点随机的力，这可能会导致其运动出现显着偏差。反馈控制器通常用于抑制干扰并回到期望轨迹。然而，运动的不确定性可能是显着的，必须考虑在轨迹规划，以避免与周围的障碍物碰撞。在USV任务的情况下，需要在船上生成状态转移概率，以计算可以处理动态变化的USV参数和环境（例如，由于燃料消耗而改变的船惯性张量、由于水密度的变化而引起的阻尼的变化、海况的变化等）。本文所报道的六自由度动力学仿真是基于势流理论。为了提高仿真计算的性能，我们还提出了一种基于时间相干性的模型简化算法及其GPU实现。使用本文中讨论的技术，我们能够在不到10分钟内计算状态转移概率。计算的转移概率随后用于基于随机动态规划的方法来解决MDP获得轨迹计划。使用这种方法，我们能够生成动态可行的轨迹USVs表现出安全的行为，在高海况附近的静态障碍物。
This paper describes GPU based algorithms to compute state transition models for unmanned surface vehicles (USVs) using 6 degree of freedom (DOF) dynamics simulations of vehicle–wave interaction. A state transition model is a key component of the Markov Decision Process (MDP), which is a natural framework to formulate the problem of trajectory planning under motion uncertainty. The USV trajectory planning problem is characterized by the presence of large and somewhat stochastic forces due to ocean waves, which can cause significant deviations in their motion. Feedback controllers are often employed to reject disturbances and get back on the desired trajectory. However, the motion uncertainty can be significant and must be considered in the trajectory planning to avoid collisions with the surrounding obstacles. In case of USV missions, state transition probabilities need to be generated on-board, to compute trajectory plans that can handle dynamically changing USV parameters and environment (e.g., changing boat inertia tensor due to fuel consumption, variations in damping due to changes in water density, variations in sea-state, etc.). The 6 DOF dynamics simulations reported in this paper are based on potential flow theory. We also present a model simplification algorithm based on temporal coherence and its GPU implementation to accelerate simulation computation performance. Using the techniques discussed in this paper we were able to compute state transition probabilities in less than 10 min. Computed transition probabilities are subsequently used in a stochastic dynamic programming based approach to solve the MDP to obtain trajectory plan. Using this approach, we are able to generate dynamically feasible trajectories for USVs that exhibit safe behaviors in high sea-states in the vicinity of static obstacles.