A key component of robotic path planning isudensuring that one can reliably navigate a vehicle to a desiredudlocation. In addition, when the features of interest are dynamicudand move with oceanic currents, vehicle speed plays an importantudrole in the planning exercise to ensure that vehiclesudare in the right place at the right time. Aquatic robot design isudmoving towards utilizing the environment for propulsion ratherudthan traditional motors and propellers. These new vehicles areudable to realize significantly increased endurance, however theudmission planning problem, in turn, becomes more difficult asudthe vehicle velocity is not directly controllable. In this paper,udwe examine Gaussian process models applied to existing waveudmodel data to predict the behavior, i.e., velocity, of a WaveudGlider Autonomous Surface Vehicle. Using training data fromudan on-board sensor and forecasting with the WAVEWATCH IIIudmodel, our probabilistic regression models created an effectiveudmethod for forecasting WG velocity.
展开▼
机译：机器人路径规划的关键组成部分是确保人们可以可靠地将车辆导航到期望的位置。此外，当感兴趣的特征随洋流动态变化时，车速在规划工作中起着重要作用，以确保车辆在正确的时间正确地行驶。与传统的电动机和螺旋桨相比，水上机器人的设计正朝着利用环境推进的方向发展。这些新的车辆有可能实现耐久性的显着提高，但是，由于无法直接控制车速，因此排放计划问题变得更加困难。在本文中， udwe研究了应用于现有wave udmodel数据的高斯过程模型，以预测Wave udGlider自主地面飞行器的行为，即速度。利用来自 udan车载传感器的训练数据并使用WAVEWATCH III udmodel进行预测，我们的概率回归模型创建了一种有效的 udmethod来预测白炽速。
展开▼
