基本信息:
- 专利标题： DRIVING PREDICTION WITH A DEEP NEURAL NETWORK
- 专利标题（中）：一种深度神经网络的驱动预测
- 申请号：EP17160823.5 申请日：2017-03-14
- 公开(公告)号：EP3219564A1 公开(公告)日：2017-09-20
- 发明人： TSISHKOU, Dzmitry , BENDAHAN, Rémy
- 申请人： IMRA Europe S.A.S.
- 申请人地址： 220, Rue Albert Caquot 06904 Sophia Antipolis FR
- 专利权人： IMRA Europe S.A.S.
- 当前专利权人： IMRA Europe S.A.S.
- 当前专利权人地址： 220, Rue Albert Caquot 06904 Sophia Antipolis FR
- 代理机构： Vigand, Philippe
- 优先权： DE102016204176 20160314
- 主分类号： B60W30/10
- IPC分类号： B60W30/10 ; B60W50/14 ; B60W10/04 ; B60W10/184 ; B60W10/20 ; G05D1/02 ; G01C21/32
摘要：
Method for assisting a driver to drive a vehicle or for driving a vehicle, the method comprising a learning phase and a driving phase, the learning phase comprising the steps of:
- taking a series of pictures of followed path from a driver's point of view, during at least one journey where a vehicle is driven by a human driver,
- recording during said at least one journey driving conditions including at least a steering angle and/or a vehicle acceleration,
- for each picture taken in the series of pictures, linking to said picture the driving conditions recorded at the instant when said picture was taken,
- supplying to a deep neural network said pictures and the linked driving conditions, to associate:
- images and/or driving conditions at a given instant Ti and/or at a prior time,
with
- driving conditions at a later instant than said given instant Ti.
摘要（中）：
- taking a series of pictures of followed path from a driver's point of view, during at least one journey where a vehicle is driven by a human driver,
- recording during said at least one journey driving conditions including at least a steering angle and/or a vehicle acceleration,
- for each picture taken in the series of pictures, linking to said picture the driving conditions recorded at the instant when said picture was taken,
- supplying to a deep neural network said pictures and the linked driving conditions, to associate:
- images and/or driving conditions at a given instant Ti and/or at a prior time,
with
- driving conditions at a later instant than said given instant Ti.
1。一种用于辅助驾驶员驾驶车辆或驾驶车辆的方法，所述方法包括学习阶段和驾驶阶段，所述学习阶段包括以下步骤： - 从驾驶员的角度拍摄一系列跟随路径的画面， 在由人驾驶员驾驶车辆的至少一个行程期间， - 在所述至少一次行程驾驶条件期间记录至少包括转向角和/或车辆加速度， - 对于在所述一系列图片中拍摄的每张照片，链接 向所述图片显示在拍摄所述图片时记录的驾驶状况， - 向深度神经网络提供所述图片和所链接的驾驶条件，以将： - 在给定时刻Ti和/或在给定时刻Ti处的图像和/或驾驶状况 在先前的时间，驾驶条件在比所述给定时刻Ti更晚的时刻。
公开/授权文献:
- EP3219564B1 DRIVING PREDICTION WITH A DEEP NEURAL NETWORK 公开/授权日：2018-12-05
IPC结构图谱:
| B | 作业；运输 |
| --B60 | 一般车辆 |
| ----B60W | 不同类型或不同功能的车辆子系统的联合控制；专门适用于混合动力车辆的控制系统；不与某一特定子系统的控制相关联的道路车辆驾驶控制系统 |
| ------B60W30/00 | 不与某一特定子系统的控制相关联的道路车辆驾驶控制系统的使用 |
| --------B60W30/10 | .路线保持 |