Automated segmentation and activity identification for intelligent vehicular systems

车辆运输是现代基础设施和现代经济的关键。改善我们的 了解驾驶员在交通状况下的行为和情景感知可以改善 驾驶员安全、燃料使用和降低成本。当前用于分析驾驶行为的系统依赖于习惯 车载传感系统，它将范围限制在车辆的一小部分。这项建议的目的是 开发使用车载和移动传感器(如智能手机)的驾驶员行为分析算法。 该提案将开发用于离线和在线分割和驾驶识别的算法 来自移动设备数据的行为。离线分析可用于诸如基于使用的应用 保险、驾驶员培训和车队情报。在线分析将为互动打开新的可能性 与用户一起，提供了更直观、更方便的界面，实现了即时和交互 反馈。

Vehicular transportation is a key linchpin of modern infrastructure and the modern economy. Improving our understanding of driver behaviour and situational awareness in traffic conditions can lead to improvements in driver safety, fuel usage and decrease costs. Current systems for analysis of driving behaviour rely on custom in-vehicle sensing systems, which limit the scope to a small subset of vehicles. The aim of this proposal is to develop algorithms for driver behaviour analysis using both on-board and mobile sensors such as smartphones. The proposal would develop algorithms for both off-line and on-line segmentation and identification of driving behaviour from mobile device data. Off-line analysis can be employed for applications such as usage-based insurance, driver training and fleet intelligence. On-line analysis would open new possibilities for interaction with the user, providing a more intuitive and convenient interface, and enabling immediate and interactive feedback.