Smart vehicles, capable of exchanging necessary information with each other and transportation infrastructure, have gained much attraction in automotive research. The data are gathered from location, velocity, acceleration and heading measurements, allowing vehicles to make smart decisions regarding safety and other applications. Specifically, accurate prediction of vehicle location measurement is considered crucial for making prompt decisions in emergency situations. These connected vehicles, equipped with advanced technologies, tend to improve driver safety and mobility radically. Still, most of the current vehicular safety applications rely on sensor measurements and uncertainty associated with them. In this paper, we have calculated the uncertainty of measurement for the deep learning-based long short-term memory model developed to estimate future location for smart vehicles. The prediction is effectively performed by exploiting the data retrieved from the past trajectory of the vehicle. Most of the available models, designed to predict a vehicle's location, do not provide any information about uncertainty in their measurements. This research aims to evaluate the uncertainty of measurement in prediction error and validation loss related to location prediction, enabling the system to make reliable decisions in the context of safety applications.

能够相互交换必要信息和交通基础设施的智能汽车在汽车研究中获得了很大的吸引力。从位置、速度、加速度和航向测量中收集数据，使车辆能够就安全和其他应用做出明智的决策。具体而言，车辆位置测量的准确预测对于在紧急情况下做出及时决策至关重要。这些配备先进技术的联网车辆往往会从根本上提高驾驶员的安全性和机动性。尽管如此，大多数当前的车辆安全应用依赖于传感器测量和与之相关的不确定性。在本文中，我们已经计算了基于深度学习的长短期记忆模型的测量不确定性，该模型旨在估计智能车辆的未来位置。通过利用从车辆过去轨迹中检索到的数据，可以有效地执行预测。大多数用于预测车辆位置的可用模型不提供有关其测量不确定性的任何信息。本研究旨在评估与位置预测相关的预测误差和验证损失的测量不确定性，使系统能够在安全应用的背景下做出可靠的决策。s 位置，不提供有关其测量不确定度的任何信息。本研究旨在评估与位置预测相关的预测误差和验证损失的测量不确定性，使系统能够在安全应用的背景下做出可靠的决策。s 位置，不提供有关其测量不确定度的任何信息。本研究旨在评估与位置预测相关的预测误差和验证损失的测量不确定性，使系统能够在安全应用的背景下做出可靠的决策。