第三期 预测——2.输入和输出

 

 

预测的输入和输出

预测模块使用来自传感器融合的地图和数据来生成关于所有其他动态对象可能做的预测。为了更清楚地说明，我们来看一个预测输入和输出的例子（json格式）。

示例输入 - 传感器融合

{
    "timestamp" : 34512.21,
    "vehicles" : [ { "id" : 0, "x" : -10.0, "y" : 8.1, "v_x" : 8.0, "v_y" : 0.0, "sigma_x" : 0.031, "sigma_y" : 0.040, "sigma_v_x" : 0.12, "sigma_v_y" : 0.03, }, { "id" : 1, "x" : 10.0, "y" : 12.1, "v_x" : -8.0, "v_y" : 0.0, "sigma_x" : 0.031, "sigma_y" : 0.040, "sigma_v_x" : 0.12, "sigma_v_y" : 0.03, }, ] } 

示例输出

{
    "timestamp" : 34512.21,
    "vehicles" : [ { "id" : 0, "length": 3.4, "width" : 1.5, "predictions" : [ { "probability" : 0.781, "trajectory" : [ { "x": -10.0, "y": 8.1, "yaw": 0.0, "timestamp": 34512.71 }, { "x": -6.0, "y": 8.1, "yaw": 0.0, "timestamp": 34513.21 }, { "x": -2.0, "y": 8.1, "yaw": 0.0, "timestamp": 34513.71 }, { "x": 2.0, "y": 8.1, "yaw": 0.0, "timestamp": 34514.21 }, { "x": 6.0, "y": 8.1, "yaw": 0.0, "timestamp": 34514.71 }, { "x": 10.0, "y": 8.1, "yaw": 0.0, "timestamp": 34515.21 }, ] }, { "probability" : 0.219, "trajectory" : [ { "x": -10.0, "y": 8.1, "yaw": 0.0, "timestamp": 34512.71 }, { "x": -7.0, "y": 7.5, "yaw": -5.2, "timestamp": 34513.21 }, { "x": -4.0, "y": 6.1, "yaw": -32.0, "timestamp": 34513.71 }, { "x": -3.0, "y": 4.1, "yaw": -73.2, "timestamp": 34514.21 }, { "x": -2.0, "y": 1.2, "yaw": -90.0, "timestamp": 34514.71 }, { "x": -2.0, "y":-2.8, "yaw": -90.0, "timestamp": 34515.21 }, ] } ] }, { "id" : 1, "length": 3.4, "width" : 1.5, "predictions" : [ { "probability" : 1.0, "trajectory" : [ { "x": 10.0, "y": 12.1, "yaw": -180.0, "timestamp": 34512.71 }, { "x": 6.0, "y": 12.1, "yaw": -180.0, "timestamp": 34513.21 }, { "x": 2.0, "y": 12.1, "yaw": -180.0, "timestamp": 34513.71 }, { "x": -2.0, "y": 12.1, "yaw": -180.0, "timestamp": 34514.21 }, { "x": -6.0, "y": 12.1, "yaw": -180.0, "timestamp": 34514.71 }, { "x": -10.0, "y": 12.1, "yaw": -180.0, "timestamp": 34515.21 } ] } ] } ] } 

笔记

1. 这里显示的预测轨迹仅延伸几秒钟。实际上，我们所做的预测可以延伸到10-20秒的范围。
2. 显示的轨迹具有0.5秒的分辨率。实际上，我们会产生稍微更精细的预测。
3. 这个例子只显示，vehicles但实际上我们也会为所有动态对象产生预测。

 

练习题

左侧车辆有多少种可能的轨迹（id为0）？

• 0

• 1

• 2

• 3+

提交