Apollo教程笔记5——Prediction

文章目录

• 1.Introduction to Prediction
• 2.Sebastian's Take
• 3.Prediction Approaches
• 4.Lane-Sequence-Based Prediction
• 5.Obstacle Status
• 6.Predict Target Lane
• 7.Recurrent Neural Networks
• 8.RNNs for Target Lane Prediction
• 9.Trajectory Generation
• 10.Summary

1.Introduction to Prediction

实时性, 算法的延迟越短越好
准确性,

2.Sebastian’s Take

• model-based
• data-driven

3.Prediction Approaches

4.Lane-Sequence-Based Prediction

Apollo 提供了一种叫基于车道序列的方法

5.Obstacle Status

6.Predict Target Lane

将预测问题简化为选择问题

7.Recurrent Neural Networks

神经网络从数据中学习的方式叫做后向传播

8.RNNs for Target Lane Prediction

9.Trajectory Generation

轨迹生成是预测的最后一步

10.Summary