主题介绍
本视频通过一个简单的AEB例子阐述AnsysAutonomy自动驾驶解决方案。该自动驾驶工具链作为自动驾驶安全第一(SaFAD)的一种实现,结合了设计安全和验证安全。Ansys Autonomy通过使用基于模型的系统和软件工程(MBSE)实践,达成设计实现安全性,其中系统被设计为系统和软件模型,实施全面的可追溯性、强大的分析和验证技术,以及经过认证的从模型到代码的自动生成。Ansys Autonomy通过其全面的仿真工具链实现了安全性验证,该工具链可以进行成千上万的大规模仿真运行,在传感器模拟的保真度方面具有高度灵活性,从理想传感器到随机传感器,再到基于物理的传感器,而且仿真可以和道路驾驶数据和驾驶数据分析相联系。我们通过一个AEB的例子说明设计模型与仿真验证的结合,展示了Ansys Autonomy自动驾驶解决方案。
如有任何问题请点击以下链接进入答疑室与我们的技术专家进行交流互动
https://v.ansys.com.cn/live/48830827
演讲人介绍
Bernard Dion Ansys系统事业部CTO
(该演讲为Ansys Simulation World 虚拟大会视频)
-
00:00:00.00 - 00:00:05.10 9
早上 下午好 我是
-
00:00:05.10 - 00:00:06.12 12
Bernard Dion
-
00:00:06.24 - 00:00:11.34 14
Ansys系统部门首席技术官
-
00:00:11.34 - 00:00:11.57 8
我准备给大家讲解
-
00:00:11.68 - 00:00:16.45 18
实践中的Ansys Autonomy
-
00:00:16.56 - 00:00:16.79 14
Sandeep Sovani
-
00:00:16.79 - 00:00:21.13 10
之前已经给大家介绍了
-
00:00:21.13 - 00:00:21.61 3
什么是
-
00:00:21.70 - 00:00:26.04 16
Ansys Autonomy以及
-
00:00:26.04 - 00:00:26.52 7
我们的目标市场
-
00:00:26.62 - 00:00:30.96 12
该产品和解决方案是为了让
-
00:00:30.96 - 00:00:33.08 6
企业能够高效
-
00:00:33.08 - 00:00:36.48 13
研发基于设计与验证安全性的
-
00:00:36.48 - 00:00:38.15 9
自动驾驶汽车和组件
-
00:00:38.23 - 00:00:41.64 2
因此
-
00:00:41.64 - 00:00:41.71 5
我想给大家
-
00:00:41.79 - 00:00:45.12 9
讲一下它的实践意义
-
00:00:45.12 - 00:00:48.51 6
首先来看一份
-
00:00:48.51 - 00:00:48.88 3
白皮书
-
00:00:48.96 - 00:00:52.35 13
名字叫作 自动驾驶安全第一
-
00:00:52.35 - 00:00:53.70 4
它由众多
-
00:00:53.78 - 00:00:57.17 10
汽车公司与供应商编制
-
00:00:57.17 - 00:00:58.60 7
它分为两大部分
-
00:00:58.60 - 00:01:02.86 7
其中有设计安全
-
00:01:02.86 - 00:01:03.43 11
我们解决该问题的方式是
-
00:01:03.53 - 00:01:07.79 13
基于MBSE模型的系统工程
-
00:01:07.79 - 00:01:09.78 7
优秀实践与工具
-
00:01:09.88 - 00:01:14.14 7
另一方面是实现
-
00:01:14.14 - 00:01:15.09 4
验证安全
-
00:01:15.09 - 00:01:19.47 10
依靠大量的仿真与路试
-
00:01:19.47 - 00:01:21.61 4
道路驾驶
-
00:01:21.71 - 00:01:26.09 13
也就是场景与道路驾驶相结合
-
00:01:26.09 - 00:01:26.77 2
当然
-
00:01:26.87 - 00:01:31.24 10
明确目标之后仿真要比
-
00:01:31.24 - 00:01:31.93 7
道路驾驶多得多
-
00:01:31.93 - 00:01:35.53 10
这个解决方案的基础是
-
00:01:35.53 - 00:01:37.61 10
将不同学科融合在一起
-
00:01:37.69 - 00:01:41.30 8
中间是系统安全性
-
00:01:41.30 - 00:01:42.18 2
然后
-
00:01:42.26 - 00:01:45.87 7
左边是很重要的
-
00:01:45.87 - 00:01:46.59 9
自动驾驶系统传感器
-
00:01:46.59 - 00:01:50.04 10
右边是我们必须开发的
-
00:01:50.12 - 00:01:53.73 6
控制器与AI
-
00:01:53.73 - 00:01:53.81 3
上下是
-
00:01:53.89 - 00:01:57.49 7
系统仿真概念与
-
00:01:57.49 - 00:01:59.10 6
虚拟现实体验
-
00:01:59.10 - 00:02:02.74 21
这是在Ansys Autonomy解决方案
-
00:02:02.82 - 00:02:06.63 11
的组合产品中的所有部件
-
00:02:06.63 - 00:02:08.15 4
都需要的
-
00:02:08.24 - 00:02:12.05 15
大家看到面向安全的MEDINI
-
00:02:12.05 - 00:02:13.48 16
面向传感器的SPEOS和HFSS
-
00:02:13.48 - 00:02:17.39 15
面向嵌入式软件的SCADE以及
-
00:02:17.39 - 00:02:18.35 19
面向系统仿真的Twin Builder
-
00:02:18.43 - 00:02:22.34 15
和VRXPERIENCE 此外
-
00:02:22.34 - 00:02:23.21 6
我们需要依靠
-
00:02:23.30 - 00:02:27.20 6
此处您看到的
-
00:02:27.20 - 00:02:27.29 5
V循环模板
-
00:02:27.29 - 00:02:31.00 8
然后完成各个步骤
-
00:02:31.00 - 00:02:31.91 3
首先从
-
00:02:31.99 - 00:02:35.70 10
ODD与功能定义开始
-
00:02:35.70 - 00:02:36.61 7
在第一轮循环中
-
00:02:36.69 - 00:02:40.41 6
定义功能架构
-
00:02:40.41 - 00:02:41.73 2
同时
-
00:02:41.73 - 00:02:45.28 2
分析
-
00:02:45.28 - 00:02:46.70 5
功能安全性
-
00:02:46.78 - 00:02:50.33 5
一旦构建了
-
00:02:50.33 - 00:02:50.41 4
模型迭代
-
00:02:50.49 - 00:02:54.04 5
我们就可以
-
00:02:54.04 - 00:02:54.83 7
通过为最终系统
-
00:02:54.83 - 00:02:58.37 8
分配功能架构组件
-
00:02:58.37 - 00:03:00.02 3
来构建
-
00:03:00.10 - 00:03:03.64 4
物理架构
-
00:03:03.64 - 00:03:05.21 10
在完成系统的描述之后
-
00:03:05.29 - 00:03:08.83 6
我们可以着手
-
00:03:08.83 - 00:03:09.14 4
构建组件
-
00:03:09.14 - 00:03:12.77 12
此处我们看到的是软件部分
-
00:03:12.77 - 00:03:13.18 5
我们在这里
-
00:03:13.26 - 00:03:16.89 11
将提供工具和传感器部件
-
00:03:16.89 - 00:03:18.02 7
同样可以在这里
-
00:03:18.10 - 00:03:21.74 11
提供包括仿真在内的工具
-
00:03:21.74 - 00:03:23.27 6
以便进行验证
-
00:03:23.27 - 00:03:26.74 13
然后我们看V循环的上升部分
-
00:03:26.82 - 00:03:30.46 8
其中我们需要处理
-
00:03:30.46 - 00:03:32.24 4
各级认证
-
00:03:32.32 - 00:03:35.95 11
包括模型在环 软件在环
-
00:03:35.95 - 00:03:36.36 4
硬件在环
-
00:03:36.36 - 00:03:39.74 7
最后我们将进行
-
00:03:39.74 - 00:03:40.64 4
系统验证
-
00:03:40.71 - 00:03:44.09 7
并构建安全案例
-
00:03:44.09 - 00:03:44.77 11
其中包含证明系统安全性
-
00:03:44.85 - 00:03:48.23 5
的所有证据
-
00:03:48.23 - 00:03:48.38 3
到这里
-
00:03:48.38 - 00:03:51.71 6
并非就结束了
-
00:03:51.71 - 00:03:51.86 7
因为通常情况下
-
00:03:51.94 - 00:03:55.27 10
企业希望实现项目循环
-
00:03:55.27 - 00:03:55.72 2
希望
-
00:03:55.79 - 00:03:59.13 14
重复利用项目中已经完成的工作
-
00:03:59.13 - 00:04:00.24 6
重复利用组件
-
00:04:00.24 - 00:04:03.58 11
这样就能继续下一个循环
-
00:04:03.58 - 00:04:04.03 2
希望
-
00:04:04.10 - 00:04:06.85 14
能够重复利用之前项目已经完成
-
00:04:06.92 - 00:04:10.26 5
的众多成果
-
00:04:10.26 - 00:04:10.48 3
这就是
-
00:04:10.48 - 00:04:14.67 14
我们针对面向ADAS/AD的
-
00:04:14.67 - 00:04:15.13 15
MBSE工作流程提出的典型设置
-
00:04:15.22 - 00:04:19.40 4
它的起点
-
00:04:19.40 - 00:04:20.33 1
是
-
00:04:20.43 - 00:04:24.61 8
ODD 功能定义
-
00:04:24.61 - 00:04:25.44 3
然后是
-
00:04:25.44 - 00:04:29.63 11
构建大家看到位于中间的
-
00:04:29.63 - 00:04:30.09 4
系统模型
-
00:04:30.18 - 00:04:34.37 12
这是一个MEDINI工具
-
00:04:34.37 - 00:04:34.74 6
它为系统模型
-
00:04:34.83 - 00:04:39.01 9
提供了进行安全分析
-
00:04:39.01 - 00:04:39.48 7
所需的所有部件
-
00:04:39.48 - 00:04:42.99 17
故障分析等一般操作所需要的功能安全
-
00:04:42.99 - 00:04:44.79 4
网络安全
-
00:04:44.87 - 00:04:48.38 14
以及从初始需求提出附加要求的
-
00:04:48.38 - 00:04:49.00 6
HAZOP与
-
00:04:49.08 - 00:04:52.59 4
FMEA
-
00:04:52.59 - 00:04:54.54 7
由于已经完成了
-
00:04:54.54 - 00:04:57.82 4
这些分析
-
00:04:57.82 - 00:04:58.48 4
我们将在
-
00:04:58.55 - 00:05:01.83 11
演示中举一个很小的例子
-
00:05:01.83 - 00:05:03.37 7
也就是紧急制动
-
00:05:03.44 - 00:05:06.72 8
我们可以假设这是
-
00:05:06.72 - 00:05:07.59 4
初始架构
-
00:05:07.59 - 00:05:10.88 14
我们可以在左侧看到雷达传感器
-
00:05:10.96 - 00:05:14.32 10
然后是AEB核心功能
-
00:05:14.32 - 00:05:14.85 9
根据距离碰撞的时间
-
00:05:14.92 - 00:05:18.28 10
可能会有实施紧急制动
-
00:05:18.28 - 00:05:19.33 3
的决策
-
00:05:19.33 - 00:05:23.62 9
它进入制动系统输出
-
00:05:23.62 - 00:05:24.19 4
现在来作
-
00:05:24.29 - 00:05:25.72 5
进一步分析
-
00:05:25.72 - 00:05:29.59 12
我们可能需要探讨相关规格
-
00:05:29.59 - 00:05:30.79 2
例如
-
00:05:30.88 - 00:05:31.57 3
在左侧
-
00:05:31.57 - 00:05:36.04 10
我们可以看到开始建模
-
00:05:36.04 - 00:05:36.44 3
场景时
-
00:05:36.54 - 00:05:41.01 7
我们有可能撞击
-
00:05:41.01 - 00:05:41.21 4
绿色汽车
-
00:05:41.31 - 00:05:44.79 11
也有可能不撞击它 因为
-
00:05:44.79 - 00:05:48.22 7
如果有司机跟在
-
00:05:48.22 - 00:05:48.98 4
道路后面
-
00:05:49.06 - 00:05:52.42 8
这就不会带来问题
-
00:05:52.49 - 00:05:55.93 4
这预示着
-
00:05:55.93 - 00:05:56.08 10
需求分析和初始架构中
-
00:05:56.08 - 00:05:59.51 7
遗漏了什么东西
-
00:05:59.51 - 00:06:01.27 5
这会让我们
-
00:06:01.34 - 00:06:04.77 9
不得不考虑这个问题
-
00:06:04.77 - 00:06:05.31 5
考虑该问题
-
00:06:05.38 - 00:06:08.81 13
这会极大地促进教学因果分析
-
00:06:08.81 - 00:06:10.11 8
也就是FTA分析
-
00:06:10.11 - 00:06:13.78 11
其中大家在这里可以看出
-
00:06:13.87 - 00:06:15.54 11
如果我们在研究非法制动
-
00:06:15.54 - 00:06:18.90 16
那么目标之一可能是与自我车道交叉
-
00:06:18.98 - 00:06:22.12 3
的对象
-
00:06:22.12 - 00:06:25.42 5
虽然这不是
-
00:06:25.50 - 00:06:28.88 9
我们之前看到的情况
-
00:06:28.88 - 00:06:29.33 8
但是这仍然会导致
-
00:06:29.41 - 00:06:32.79 7
修改大家所知的
-
00:06:32.79 - 00:06:33.70 4
通道架构
-
00:06:33.70 - 00:06:37.45 9
我们具备通道分隔符
-
00:06:37.45 - 00:06:37.95 4
而且我们
-
00:06:38.04 - 00:06:39.71 6
需要考虑车道
-
00:06:39.71 - 00:06:43.13 13
一切取决于是否是完全自动化
-
00:06:43.13 - 00:06:43.44 2
但是
-
00:06:43.51 - 00:06:46.71 12
任何情况下处理的依据都是
-
00:06:46.78 - 00:06:50.21 4
车道信息
-
00:06:50.21 - 00:06:50.89 9
然后我们对哪些用于
-
00:06:50.89 - 00:06:54.32 6
触发紧急制动
-
00:06:54.32 - 00:06:55.00 4
做出决策
-
00:06:55.08 - 00:06:58.50 11
因此这就是该架构的迭代
-
00:06:58.50 - 00:06:59.11 5
它可以解决
-
00:06:59.19 - 00:07:02.61 10
我们在初始需求分析中
-
00:07:02.61 - 00:07:02.84 6
所确定的问题
-
00:07:02.84 - 00:07:06.40 2
之一
-
00:07:06.40 - 00:07:07.66 8
现在我们继续进行
-
00:07:07.74 - 00:07:11.30 4
仿真部分
-
00:07:11.30 - 00:07:12.09 10
那么这是对我们提供的
-
00:07:12.17 - 00:07:15.72 9
仿真内容的深层描述
-
00:07:15.72 - 00:07:16.04 6
我们可以看出
-
00:07:16.04 - 00:07:20.32 7
它从ODD定义
-
00:07:20.32 - 00:07:20.42 2
开始
-
00:07:20.42 - 00:07:23.70 10
它包含场景创建与变化
-
00:07:23.70 - 00:07:25.09 4
测试计划
-
00:07:25.16 - 00:07:28.45 8
以及执行闭环仿真
-
00:07:28.45 - 00:07:28.96 4
其中我们
-
00:07:29.03 - 00:07:32.31 10
有环境模型与场景模型
-
00:07:32.31 - 00:07:32.90 4
我们准备
-
00:07:32.90 - 00:07:36.19 10
在这里测试来自外部的
-
00:07:36.19 - 00:07:36.62 4
AD功能
-
00:07:36.70 - 00:07:39.98 4
然后研究
-
00:07:39.98 - 00:07:40.79 4
预期结果
-
00:07:40.86 - 00:07:44.15 6
这只是一部分
-
00:07:44.15 - 00:07:44.44 2
此外
-
00:07:44.44 - 00:07:47.40 12
我们在这里还看到其它部分
-
00:07:47.47 - 00:07:50.50 14
因此首先是左侧的驾驶数据以及
-
00:07:50.57 - 00:07:53.82 13
如何将其充分利用 我们准备
-
00:07:53.82 - 00:07:54.26 4
开展分析
-
00:07:54.26 - 00:07:57.45 8
我们通过驾驶分析
-
00:07:57.45 - 00:07:58.23 12
有可能会发现我们在道路上
-
00:07:58.30 - 00:08:01.49 7
看到的情景并非
-
00:08:01.49 - 00:08:02.27 7
场景中描述那样
-
00:08:02.35 - 00:08:05.47 9
在这种情况下就需要
-
00:08:05.47 - 00:08:08.66 14
再次考虑相关场景 而且有可能
-
00:08:08.66 - 00:08:09.79 8
需要添加仿真场景
-
00:08:09.86 - 00:08:13.05 11
然后把它们纳入测试计划
-
00:08:13.05 - 00:08:13.62 5
另一个问题
-
00:08:13.69 - 00:08:16.88 15
是关于仿真 或者说一个并非难题
-
00:08:16.88 - 00:08:17.59 3
的疑问
-
00:08:17.59 - 00:08:20.85 16
涉及工具链验证 我们是否可以信任
-
00:08:20.85 - 00:08:21.72 4
仿真以及
-
00:08:21.79 - 00:08:25.05 11
相关结果在提供帮助方面
-
00:08:25.05 - 00:08:25.84 10
是否足够接近现实情况
-
00:08:25.92 - 00:08:29.17 2
为了
-
00:08:29.17 - 00:08:29.24 9
考虑并解决这个问题
-
00:08:29.24 - 00:08:32.69 4
我们同样
-
00:08:32.69 - 00:08:32.92 9
具备模块工具链验证
-
00:08:33.00 - 00:08:36.44 2
它从
-
00:08:36.44 - 00:08:37.44 6
驾驶数据开始
-
00:08:37.52 - 00:08:40.81 10
我们利用这种驾驶数据
-
00:08:40.81 - 00:08:44.25 6
重新创建场景
-
00:08:44.25 - 00:08:45.10 10
并且把拥有的结果以及
-
00:08:45.17 - 00:08:48.62 4
驾驶数据
-
00:08:48.62 - 00:08:49.00 8
与仿真的相同场景
-
00:08:49.08 - 00:08:52.52 3
的结果
-
00:08:52.52 - 00:08:52.98 4
进行比较
-
00:08:52.98 - 00:08:56.51 12
以确定它们是否如我们需要
-
00:08:56.60 - 00:08:57.63 7
的那样接近现实
-
00:08:57.63 - 00:09:01.14 9
最后但同样重要的是
-
00:09:01.14 - 00:09:01.37 4
数据沿袭
-
00:09:01.45 - 00:09:04.97 11
我们需要始终管理数据并
-
00:09:04.97 - 00:09:05.28 2
跟踪
-
00:09:05.36 - 00:09:08.87 11
一切操作以及仿真过程中
-
00:09:08.87 - 00:09:10.12 4
所执行的
-
00:09:10.12 - 00:09:13.63 7
所有元素之间的
-
00:09:13.63 - 00:09:14.33 3
相依性
-
00:09:14.41 - 00:09:17.93 2
为了
-
00:09:17.93 - 00:09:18.39 4
深入介绍
-
00:09:18.47 - 00:09:21.98 10
第一点是建模现实世界
-
00:09:21.98 - 00:09:23.16 6
在此可以看到
-
00:09:23.16 - 00:09:26.75 10
我们可以创建一个场景
-
00:09:26.75 - 00:09:26.83 3
它源于
-
00:09:26.91 - 00:09:30.51 8
HD地图信息以及
-
00:09:30.51 - 00:09:31.39 7
用于创建场景的
-
00:09:31.47 - 00:09:32.43 6
其他类型信息
-
00:09:32.43 - 00:09:36.70 13
然后我们需要创建传感器仿真
-
00:09:36.70 - 00:09:37.08 2
为了
-
00:09:37.18 - 00:09:41.45 7
给大家举例说明
-
00:09:41.45 - 00:09:41.64 5
这里展示了
-
00:09:41.74 - 00:09:46.01 10
仿真精度可以根据需要
-
00:09:46.01 - 00:09:46.68 8
达到非常高的水平
-
00:09:46.68 - 00:09:50.92 6
这里是摄像头
-
00:09:50.92 - 00:09:51.58 5
传感器示例
-
00:09:51.68 - 00:09:55.92 9
传感器插入前灯内部
-
00:09:55.92 - 00:09:57.06 7
这里的仿真研究
-
00:09:57.15 - 00:10:01.40 6
内容是雨滴对
-
00:10:01.40 - 00:10:02.44 5
前灯的影响
-
00:10:02.44 - 00:10:06.75 4
我们看到
-
00:10:06.75 - 00:10:07.23 6
白天和夜间的
-
00:10:07.32 - 00:10:11.63 4
视觉效果
-
00:10:11.63 - 00:10:12.40 3
相当差
-
00:10:12.49 - 00:10:16.80 13
另外可以看到 我们在前灯上
-
00:10:16.80 - 00:10:21.28 13
添加了疏水膜 然后视觉效果
-
00:10:21.28 - 00:10:22.37 4
改善很多
-
00:10:22.47 - 00:10:26.94 15
因此 根据需要仿真环境允许我们
-
00:10:26.94 - 00:10:28.33 10
基于传感器的逼真建模
-
00:10:28.43 - 00:10:32.90 4
研究更多
-
00:10:32.90 - 00:10:33.60 4
物理现象
-
00:10:33.60 - 00:10:37.21 9
创建仿真的最后一个
-
00:10:37.21 - 00:10:37.77 3
要点是
-
00:10:37.85 - 00:10:41.47 6
关于驾驶场景
-
00:10:41.47 - 00:10:42.27 9
因此我在此处展示的
-
00:10:42.35 - 00:10:45.96 9
是正在制作中的项目
-
00:10:45.96 - 00:10:46.53 3
因为它
-
00:10:46.53 - 00:10:50.14 13
涉及ASAM正在制订的标准
-
00:10:51.10 - 00:10:53.11 19
也就是OpenSCENARIO 2.0
-
00:10:53.11 - 00:10:54.88 10
其中包含用于创建驾驶
-
00:10:54.96 - 00:10:58.57 7
场景的高级语言
-
00:10:58.57 - 00:10:59.62 7
标准将很快问世
-
00:10:59.62 - 00:11:04.14 2
到时
-
00:11:04.14 - 00:11:05.14 7
我们就可以利用
-
00:11:05.24 - 00:11:05.65 4
这种语言
-
00:11:05.65 - 00:11:10.00 10
现在我们来看软件部分
-
00:11:10.00 - 00:11:10.38 2
同样
-
00:11:10.48 - 00:11:14.83 11
我们需要通过举例来探讨
-
00:11:14.83 - 00:11:16.28 4
软件需求
-
00:11:16.38 - 00:11:20.73 14
同时研究如何实现以及如何仿真
-
00:11:20.73 - 00:11:21.60 2
那么
-
00:11:21.60 - 00:11:25.91 10
在这里我们有软件组件
-
00:11:25.91 - 00:11:26.77 5
它依赖输入
-
00:11:26.86 - 00:11:31.17 7
也就是占用网格
-
00:11:31.17 - 00:11:32.03 12
其中我们可以看到占用区域
-
00:11:32.13 - 00:11:36.43 9
而它可能是一个物体
-
00:11:36.43 - 00:11:36.91 4
然后我们
-
00:11:36.91 - 00:11:41.06 5
了解到一种
-
00:11:41.06 - 00:11:41.98 12
通过检测碰撞威胁做出判断
-
00:11:42.07 - 00:11:46.22 5
的软件需求
-
00:11:46.22 - 00:11:48.99 15
我们可以根据占用概率 碰撞时间
-
00:11:49.08 - 00:11:53.23 7
和自主车辆速度
-
00:11:53.23 - 00:11:55.07 6
实施紧急制动
-
00:11:55.07 - 00:11:59.46 10
因此为了实现这种软件
-
00:11:59.46 - 00:12:00.24 4
我们建议
-
00:12:00.34 - 00:12:04.72 10
采用大家在这里看到的
-
00:12:04.72 - 00:12:05.60 12
基于模型的软件符号和工具
-
00:12:05.70 - 00:12:10.08 6
其中大家可以
-
00:12:10.08 - 00:12:10.67 11
用超高层次和图形化方式
-
00:12:10.67 - 00:12:14.27 6
表达相关算法
-
00:12:14.27 - 00:12:14.59 9
所以这里我不再赘述
-
00:12:14.67 - 00:12:18.27 4
不过需要
-
00:12:18.27 - 00:12:18.59 2
介绍
-
00:12:18.67 - 00:12:22.28 9
这些阵列单元的迭代
-
00:12:22.28 - 00:12:23.56 6
对象检测以及
-
00:12:23.56 - 00:12:27.08 10
了解是否应当实施制动
-
00:12:27.08 - 00:12:28.41 5
而超高层次
-
00:12:28.49 - 00:12:32.01 6
图形符号配备
-
00:12:32.01 - 00:12:33.50 10
经过认证的代码生成器
-
00:12:33.57 - 00:12:37.10 10
因此大家不必担心软件
-
00:12:37.10 - 00:12:37.80 4
代码本身
-
00:12:37.80 - 00:12:41.91 12
大家只需关心设计工作以及
-
00:12:41.91 - 00:12:43.28 4
了解设计
-
00:12:43.37 - 00:12:47.49 9
是否正确并符合要求
-
00:12:47.49 - 00:12:48.86 2
一旦
-
00:12:48.95 - 00:12:50.05 6
完成这些工作
-
00:12:50.05 - 00:12:54.36 13
您就可以开展仿真 因此我们
-
00:12:54.46 - 00:12:58.87 12
在此已经耦合了生成的代码
-
00:12:58.87 - 00:12:59.85 4
以便我们
-
00:12:59.95 - 00:13:00.93 10
结合驾驶场景的目的地
-
00:13:00.93 - 00:13:05.22 6
执行模型仿真
-
00:13:05.22 - 00:13:06.65 6
我们可以看到
-
00:13:06.75 - 00:13:11.04 8
此处车辆由于行人
-
00:13:11.04 - 00:13:11.14 5
而即将刹车
-
00:13:11.23 - 00:13:15.15 6
然后恢复行驶
-
00:13:15.15 - 00:13:18.73 15
现在我希望介绍的最后一个话题是
-
00:13:18.81 - 00:13:22.47 4
评估大家
-
00:13:22.47 - 00:13:23.28 7
对实施紧急制动
-
00:13:23.37 - 00:13:27.03 4
的置信度
-
00:13:27.03 - 00:13:28.25 12
它依赖于对众多方面的涵盖
-
00:13:28.25 - 00:13:31.83 2
首先
-
00:13:31.91 - 00:13:35.57 7
应当涵盖场景的
-
00:13:35.57 - 00:13:36.71 4
所有要求
-
00:13:36.80 - 00:13:40.46 10
而且它们必须经过测试
-
00:13:40.46 - 00:13:40.95 6
其次必须确保
-
00:13:40.95 - 00:13:44.63 12
涵盖驾驶过程中您将看到的
-
00:13:44.63 - 00:13:45.04 4
所有情况
-
00:13:45.12 - 00:13:48.80 7
否则就必须添加
-
00:13:48.80 - 00:13:48.96 7
我们所说的场景
-
00:13:49.04 - 00:13:52.48 8
我想要在此介绍的
-
00:13:52.48 - 00:13:56.21 7
最后一个例子是
-
00:13:56.21 - 00:13:56.71 7
您必须通过仿真
-
00:13:56.79 - 00:14:00.52 7
涵盖这里的所有
-
00:14:00.52 - 00:14:01.76 4
软件单元
-
00:14:01.85 - 00:14:05.00 11
这是一种用于软件部分的
-
00:14:05.00 - 00:14:08.66 6
模型覆盖工具
-
00:14:08.66 - 00:14:09.23 7
从这里可以看出
-
00:14:09.31 - 00:14:12.98 9
在我们进行所有仿真
-
00:14:12.98 - 00:14:13.55 8
的整个执行过程中
-
00:14:13.63 - 00:14:17.30 6
这个条件从未
-
00:14:17.30 - 00:14:20.47 4
设置为真
-
00:14:20.47 - 00:14:21.80 11
这意味着遗漏了什么东西
-
00:14:21.88 - 00:14:24.91 9
事实上遗漏的案例是
-
00:14:24.98 - 00:14:28.15 3
当两辆
-
00:14:28.15 - 00:14:28.50 2
汽车
-
00:14:28.50 - 00:14:31.67 5
同向行驶时
-
00:14:31.67 - 00:14:31.81 3
这表明
-
00:14:31.88 - 00:14:35.06 2
需要
-
00:14:35.06 - 00:14:35.27 2
修改
-
00:14:35.34 - 00:14:38.51 7
相关要求或模型
-
00:14:38.51 - 00:14:38.86 10
或者需要添加测试案例
-
00:14:38.86 - 00:14:42.03 4
以便能够
-
00:14:42.03 - 00:14:42.24 7
满足覆盖方面的
-
00:14:42.31 - 00:14:45.49 4
所有预期
-
00:14:45.49 - 00:14:45.70 12
这只是覆盖类型的一个示例
-
00:14:45.77 - 00:14:48.94 6
但是存在许多
-
00:14:48.94 - 00:14:49.64 5
必须实现的
-
00:14:49.64 - 00:14:52.95 6
不同覆盖类型
-
00:14:53.03 - 00:14:56.49 4
演讲到此
-
00:14:56.49 - 00:14:56.95 2
结束
-
00:14:57.03 - 00:15:00.34 10
现在我准备问几个问题
-
00:15:00.34 - 00:15:00.58 4
谢谢大家
-
00:00:00.00 - 00:00:05.10 38
Good morning Good afternoon my name is
-
00:00:05.10 - 00:00:06.12 12
Bernard Dion
-
00:00:06.24 - 00:00:11.34 34
CTO of systems at Ansys and I will
-
00:00:11.34 - 00:00:11.57 7
present
-
00:00:11.68 - 00:00:16.45 37
you now Ansys automomy in practice so
-
00:00:16.56 - 00:00:16.79 2
in
-
00:00:16.79 - 00:00:21.13 47
a previous talk by Sandeep Sovani you have seen
-
00:00:21.13 - 00:00:21.61 4
what
-
00:00:21.70 - 00:00:26.04 42
is Ansys autonomy about and what market we
-
00:00:26.04 - 00:00:26.52 7
are and
-
00:00:26.62 - 00:00:30.96 44
this product this solution is about enabling
-
00:00:30.96 - 00:00:33.08 24
companies to efficiently
-
00:00:33.08 - 00:00:36.48 42
develop autonomous vehicles and components
-
00:00:36.48 - 00:00:38.15 23
that are safe by design
-
00:00:38.23 - 00:00:41.64 42
and safe by validation so I'd like to show
-
00:00:41.64 - 00:00:41.71 3
you
-
00:00:41.79 - 00:00:45.12 44
to give you an idea in practice of what this
-
00:00:45.12 - 00:00:48.51 44
means and the starting point will be a white
-
00:00:48.51 - 00:00:48.88 5
paper
-
00:00:48.96 - 00:00:52.35 45
that is called safety first automated driving
-
00:00:52.35 - 00:00:53.70 17
created by number
-
00:00:53.78 - 00:00:57.17 44
of automotive companies and suppliers and it
-
00:00:57.17 - 00:00:58.60 19
has two fundamental
-
00:00:58.60 - 00:01:02.86 40
parts in it safe by design which we will
-
00:01:02.86 - 00:01:03.43 7
address
-
00:01:03.53 - 00:01:07.79 38
by MBSE model based system engineering
-
00:01:07.79 - 00:01:09.78 24
good practices and tools
-
00:01:09.88 - 00:01:14.14 45
and on the other hand by establishing safe by
-
00:01:14.14 - 00:01:15.09 10
validation
-
00:01:15.09 - 00:01:19.47 39
and relying on extensive simulation and
-
00:01:19.47 - 00:01:21.61 26
road driving a combination
-
00:01:21.71 - 00:01:26.09 44
of situation and road driving with of course
-
00:01:26.09 - 00:01:26.77 9
much more
-
00:01:26.87 - 00:01:31.24 47
simulation then road driving in a well targeted
-
00:01:31.24 - 00:01:31.93 6
manner
-
00:01:31.93 - 00:01:35.53 44
This solution relies on assembling different
-
00:01:35.53 - 00:01:37.61 26
disciplines together so in
-
00:01:37.69 - 00:01:41.30 45
the middle you see system safety and then for
-
00:01:41.30 - 00:01:42.18 10
autonomous
-
00:01:42.26 - 00:01:45.87 41
systems sensors are important which is on
-
00:01:45.87 - 00:01:46.59 8
the left
-
00:01:46.59 - 00:01:50.04 43
on the right we see that we have to develop
-
00:01:50.12 - 00:01:53.73 43
controls and AI and on top and bottom there
-
00:01:53.73 - 00:01:53.81 2
is
-
00:01:53.89 - 00:01:57.49 41
a notion of system simulation and virtual
-
00:01:57.49 - 00:01:59.10 23
reality experience that
-
00:01:59.10 - 00:02:02.74 43
is needed for all of these parts we do have
-
00:02:02.82 - 00:02:06.63 44
products that we are assembling in the Ansys
-
00:02:06.63 - 00:02:08.15 20
autonomy solution so
-
00:02:08.24 - 00:02:12.05 34
you see MEDINIfor safety SPEOS and
-
00:02:12.05 - 00:02:13.48 16
HFSS for sensors
-
00:02:13.48 - 00:02:17.39 44
SCADE for embedded software and twin builder
-
00:02:17.39 - 00:02:18.35 15
and VRXPERIENCE
-
00:02:18.43 - 00:02:22.34 31
for system simulation on top of
-
00:02:22.34 - 00:02:23.21 11
that we are
-
00:02:23.30 - 00:02:27.20 42
going to rely on the V cycle template that
-
00:02:27.20 - 00:02:27.29 3
you
-
00:02:27.29 - 00:02:31.00 40
see here and going through various steps
-
00:02:31.00 - 00:02:31.91 15
starting at the
-
00:02:31.99 - 00:02:35.70 45
top with ODD and feature definition and going
-
00:02:35.70 - 00:02:36.61 12
in the first
-
00:02:36.69 - 00:02:40.41 35
round of definition of a functional
-
00:02:40.41 - 00:02:41.73 25
architecture that is done
-
00:02:41.73 - 00:02:45.28 38
altogether with number of analysis for
-
00:02:45.28 - 00:02:46.70 24
functional safety and so
-
00:02:46.78 - 00:02:50.33 40
that if one there is a model's iteration
-
00:02:50.33 - 00:02:50.41 4
that
-
00:02:50.49 - 00:02:54.04 35
has been constructed we go into the
-
00:02:54.04 - 00:02:54.83 19
construction of the
-
00:02:54.83 - 00:02:58.37 37
physical architecture that is made by
-
00:02:58.37 - 00:03:00.02 28
allocating components of the
-
00:03:00.10 - 00:03:03.64 45
functional architecture onto the final system
-
00:03:03.64 - 00:03:05.21 19
and once the system
-
00:03:05.29 - 00:03:08.83 45
has been described we go into the building of
-
00:03:08.83 - 00:03:09.14 3
the
-
00:03:09.14 - 00:03:12.77 41
components so here we see a software part
-
00:03:12.77 - 00:03:13.18 8
where we
-
00:03:13.26 - 00:03:16.89 40
will provide tools and sensor part where
-
00:03:16.89 - 00:03:18.02 18
again will provide
-
00:03:18.10 - 00:03:21.74 42
tools including simulation and in order to
-
00:03:21.74 - 00:03:23.27 22
achieve validation and
-
00:03:23.27 - 00:03:26.74 43
then we go into the ascending part of the V
-
00:03:26.82 - 00:03:30.46 44
where we will proceed with various levels of
-
00:03:30.46 - 00:03:32.24 22
verification including
-
00:03:32.32 - 00:03:35.95 38
model in the loop software in the loop
-
00:03:35.95 - 00:03:36.36 11
hardware in
-
00:03:36.36 - 00:03:39.74 42
the loop and finally we're going to system
-
00:03:39.74 - 00:03:40.64 14
validation and
-
00:03:40.71 - 00:03:44.09 45
building a safety case where there is all the
-
00:03:44.09 - 00:03:44.77 8
evidence
-
00:03:44.85 - 00:03:48.23 42
that you can trust that the system is safe
-
00:03:48.23 - 00:03:48.38 4
this
-
00:03:48.38 - 00:03:51.71 45
is not the end of the story because typically
-
00:03:51.71 - 00:03:51.86 1
a
-
00:03:51.94 - 00:03:55.27 44
company wants to build on project to project
-
00:03:55.27 - 00:03:55.72 6
and to
-
00:03:55.79 - 00:03:59.13 46
reuse things that have been done in project to
-
00:03:59.13 - 00:04:00.24 16
reuse components
-
00:04:00.24 - 00:04:03.58 39
and so it continues into the next cycle
-
00:04:03.58 - 00:04:04.03 11
hoping that
-
00:04:04.10 - 00:04:06.85 37
a lot of it can be reused in what had
-
00:04:06.92 - 00:04:10.26 44
been done in the previous project so this is
-
00:04:10.26 - 00:04:10.48 3
the
-
00:04:10.48 - 00:04:14.67 40
typical setting that we propose in terms
-
00:04:14.67 - 00:04:15.13 7
of MBSE
-
00:04:15.22 - 00:04:19.40 38
workflow for ADAS/AD starting from the
-
00:04:19.40 - 00:04:20.33 13
beginning you
-
00:04:20.43 - 00:04:24.61 44
know the ODD the feature definition and then
-
00:04:24.61 - 00:04:25.44 8
there is
-
00:04:25.44 - 00:04:29.63 43
building a system model that you see in the
-
00:04:29.63 - 00:04:30.09 6
middle
-
00:04:30.18 - 00:04:34.37 44
here this is a MEDINI tool that is providing
-
00:04:34.37 - 00:04:34.74 6
around
-
00:04:34.83 - 00:04:39.01 43
system model all the elements to proceed to
-
00:04:39.01 - 00:04:39.48 6
do the
-
00:04:39.48 - 00:04:42.99 33
safety analysis functional safety
-
00:04:42.99 - 00:04:44.79 23
cybersecurity to do the
-
00:04:44.87 - 00:04:48.38 49
usual things like fault analysis HAZOP as well as
-
00:04:48.38 - 00:04:49.00 4
FMEA
-
00:04:49.08 - 00:04:52.59 43
to produce additional requirements from the
-
00:04:52.59 - 00:04:54.54 29
initial ones because of these
-
00:04:54.54 - 00:04:57.82 42
analysis that have been performed so we'll
-
00:04:57.82 - 00:04:58.48 12
take in this
-
00:04:58.55 - 00:05:01.83 44
presentation a very small example which will
-
00:05:01.83 - 00:05:03.37 21
be emergency breaking
-
00:05:03.44 - 00:05:06.72 41
and let's assume that this is the initial
-
00:05:06.72 - 00:05:07.59 15
architecture so
-
00:05:07.59 - 00:05:10.88 44
we see the radar sensor on the left and then
-
00:05:10.96 - 00:05:14.32 42
AEB core function and depending on time to
-
00:05:14.32 - 00:05:14.85 9
collision
-
00:05:14.92 - 00:05:18.28 44
there may be the decision to apply emergency
-
00:05:18.28 - 00:05:19.33 14
breaking which
-
00:05:19.33 - 00:05:23.62 45
goes onto the brake system output now looking
-
00:05:23.62 - 00:05:24.19 5
at it
-
00:05:24.29 - 00:05:25.72 15
in more details
-
00:05:25.72 - 00:05:29.59 44
We may have a discussion about specification
-
00:05:29.59 - 00:05:30.79 14
for example on
-
00:05:30.88 - 00:05:31.57 8
the left
-
00:05:31.57 - 00:05:36.04 39
What we see here when we start to model
-
00:05:36.04 - 00:05:36.44 9
scenarios
-
00:05:36.54 - 00:05:41.01 43
is that maybe we're going to hit this green
-
00:05:41.01 - 00:05:41.21 3
car
-
00:05:41.31 - 00:05:44.79 29
but maybe not because in fact
-
00:05:44.79 - 00:05:48.22 43
provided that the driver if there is driver
-
00:05:48.22 - 00:05:48.98 11
follows the
-
00:05:49.06 - 00:05:52.42 43
road then it's not going to be a problem so
-
00:05:52.49 - 00:05:55.93 43
that is a sign that something is missing in
-
00:05:55.93 - 00:05:56.08 3
the
-
00:05:56.08 - 00:05:59.51 33
requirements analysis and initial
-
00:05:59.51 - 00:06:01.27 34
architecture and this will lead to
-
00:06:01.34 - 00:06:04.77 45
you know thinking about it and thinking about
-
00:06:04.77 - 00:06:05.31 6
it can
-
00:06:05.38 - 00:06:08.81 43
be greatly help school causal analysis that
-
00:06:08.81 - 00:06:10.11 15
is FTA analysis
-
00:06:10.11 - 00:06:13.78 44
where you see here that if we are looking at
-
00:06:13.87 - 00:06:15.54 20
unjustified breaking
-
00:06:15.54 - 00:06:18.90 44
Then one of the goals may be that the object
-
00:06:18.98 - 00:06:22.12 38
reported as intersecting with ego lane
-
00:06:22.12 - 00:06:25.42 44
although it is not which is the case in what
-
00:06:25.50 - 00:06:28.88 41
we had seen previously and this will lead
-
00:06:28.88 - 00:06:29.33 9
this will
-
00:06:29.41 - 00:06:32.79 43
lead to modify architecture where we do you
-
00:06:32.79 - 00:06:33.70 13
know corridor
-
00:06:33.70 - 00:06:37.45 39
we have corridor delimiters and we will
-
00:06:37.45 - 00:06:37.95 7
we will
-
00:06:38.04 - 00:06:39.71 21
worry about the lanes
-
00:06:39.71 - 00:06:43.13 45
All depends if this is fully automated or not
-
00:06:43.13 - 00:06:43.44 3
but
-
00:06:43.51 - 00:06:46.71 36
in any case it should be handled and
-
00:06:46.78 - 00:06:50.21 38
depending on the Lane information then
-
00:06:50.21 - 00:06:50.89 8
you will
-
00:06:50.89 - 00:06:54.32 44
make a decision which will be to trigger the
-
00:06:54.32 - 00:06:55.00 9
emergency
-
00:06:55.08 - 00:06:58.50 38
braking so this is an iteration of the
-
00:06:58.50 - 00:06:59.11 12
architecture
-
00:06:59.19 - 00:07:02.61 43
that is solving one of the problems that we
-
00:07:02.61 - 00:07:02.84 4
have
-
00:07:02.84 - 00:07:06.40 42
identified in the initial requirements now
-
00:07:06.40 - 00:07:07.66 18
let's move on with
-
00:07:07.74 - 00:07:11.30 43
the simulation part so this is a high level
-
00:07:11.30 - 00:07:12.09 11
description
-
00:07:12.17 - 00:07:15.72 44
of what we provide in terms of simulation So
-
00:07:15.72 - 00:07:16.04 4
what
-
00:07:16.04 - 00:07:20.32 33
we see is that it starts from ODD
-
00:07:20.32 - 00:07:20.42 10
definition
-
00:07:20.42 - 00:07:23.70 41
He should have scenarios and variation of
-
00:07:23.70 - 00:07:25.09 22
parameters creation of
-
00:07:25.16 - 00:07:28.45 44
test plan execution of the closed loop where
-
00:07:28.45 - 00:07:28.96 2
we
-
00:07:29.03 - 00:07:32.31 44
have a model of the environment the scenario
-
00:07:32.31 - 00:07:32.90 9
and we'll
-
00:07:32.90 - 00:07:36.19 39
be testing the AD function here that is
-
00:07:36.19 - 00:07:36.62 11
coming from
-
00:07:36.70 - 00:07:39.98 45
outside and then looking at the result of the
-
00:07:39.98 - 00:07:40.79 8
expected
-
00:07:40.86 - 00:07:44.15 45
results on there so that's one part there are
-
00:07:44.15 - 00:07:44.44 5
other
-
00:07:44.44 - 00:07:47.40 41
parts that we see here so first of all on
-
00:07:47.47 - 00:07:50.50 42
the left there is drive data and how do we
-
00:07:50.57 - 00:07:53.82 37
use it well we will perform analytics
-
00:07:53.82 - 00:07:54.26 8
and from
-
00:07:54.26 - 00:07:57.45 43
this drive analytics maybe we will discover
-
00:07:57.45 - 00:07:58.23 16
that on the road
-
00:07:58.30 - 00:08:01.49 43
we see situations that are not described in
-
00:08:01.49 - 00:08:02.27 12
the scenario
-
00:08:02.35 - 00:08:05.47 44
so in that case there is need to think about
-
00:08:05.47 - 00:08:08.66 36
the scenarios again and maybe to add
-
00:08:08.66 - 00:08:09.79 24
simulation scenarios and
-
00:08:09.86 - 00:08:13.05 45
then to include them in the test plan another
-
00:08:13.05 - 00:08:13.62 7
problem
-
00:08:13.69 - 00:08:16.88 45
that there is with simulation or question not
-
00:08:16.88 - 00:08:17.59 9
a problem
-
00:08:17.59 - 00:08:20.85 42
is about toolchain validation can we trust
-
00:08:20.85 - 00:08:21.72 14
the simulation
-
00:08:21.79 - 00:08:25.05 41
are the results sufficiently close to the
-
00:08:25.05 - 00:08:25.84 14
reality in the
-
00:08:25.92 - 00:08:29.17 41
way that they are helpful and in order to
-
00:08:29.17 - 00:08:29.24 5
think
-
00:08:29.24 - 00:08:32.69 43
about this to have a decision about this we
-
00:08:32.69 - 00:08:32.92 4
also
-
00:08:33.00 - 00:08:36.44 45
have this module toolchain validation that is
-
00:08:36.44 - 00:08:37.44 13
starting from
-
00:08:37.52 - 00:08:40.81 43
drive data and on this drive data we use it
-
00:08:40.81 - 00:08:44.25 42
to reconstruct the scenario and we compare
-
00:08:44.25 - 00:08:45.10 13
the result on
-
00:08:45.17 - 00:08:48.62 45
the role that we have together with the drive
-
00:08:48.62 - 00:08:49.00 4
data
-
00:08:49.08 - 00:08:52.52 37
to the result of the same scenario in
-
00:08:52.52 - 00:08:52.98 13
simulation to
-
00:08:52.98 - 00:08:56.51 38
see if they are as close as we need to
-
00:08:56.60 - 00:08:57.63 11
the reality
-
00:08:57.63 - 00:09:01.14 39
Last but not least this comes with data
-
00:09:01.14 - 00:09:01.37 7
lineage
-
00:09:01.45 - 00:09:04.97 43
we need to manage the data forever and keep
-
00:09:04.97 - 00:09:05.28 5
track
-
00:09:05.36 - 00:09:08.87 36
of everything that has been done and
-
00:09:08.87 - 00:09:10.12 24
dependencies the tracing
-
00:09:10.12 - 00:09:13.63 41
between all the elements of what has been
-
00:09:13.63 - 00:09:14.33 12
performed in
-
00:09:14.41 - 00:09:17.93 41
these simulation activities so to go in a
-
00:09:17.93 - 00:09:18.39 10
little bit
-
00:09:18.47 - 00:09:21.98 43
more details first point modeling realistic
-
00:09:21.98 - 00:09:23.16 16
world so you see
-
00:09:23.16 - 00:09:26.75 41
here that we create a world then that can
-
00:09:26.75 - 00:09:26.83 4
come
-
00:09:26.91 - 00:09:30.51 41
from HD map information and other type of
-
00:09:30.51 - 00:09:31.39 14
information to
-
00:09:31.47 - 00:09:32.43 13
build a world
-
00:09:32.43 - 00:09:36.70 43
Then we need to build the sensor simulation
-
00:09:36.70 - 00:09:37.08 5
so to
-
00:09:37.18 - 00:09:41.45 41
give you an example here to show that the
-
00:09:41.45 - 00:09:41.64 5
level
-
00:09:41.74 - 00:09:46.01 42
of fidelity of the simulation can get very
-
00:09:46.01 - 00:09:46.68 9
high when
-
00:09:46.68 - 00:09:50.92 45
needed here we have the example of the camera
-
00:09:50.92 - 00:09:51.58 6
sensor
-
00:09:51.68 - 00:09:55.92 44
that is inserted inside headlamp and what is
-
00:09:55.92 - 00:09:57.06 12
studied here
-
00:09:57.15 - 00:10:01.40 43
simulated is the effect of rain droplets on
-
00:10:01.40 - 00:10:02.44 12
the headlamp
-
00:10:02.44 - 00:10:06.75 33
and we see that it becomes fairly
-
00:10:06.75 - 00:10:07.23 12
difficult to
-
00:10:07.32 - 00:10:11.63 43
do perception during the day and during the
-
00:10:11.63 - 00:10:12.40 9
night and
-
00:10:12.49 - 00:10:16.80 45
what we see is that we add a hydrophobic film
-
00:10:16.80 - 00:10:21.28 43
on the headlamp then the perception becomes
-
00:10:21.28 - 00:10:22.37 14
much easier so
-
00:10:22.47 - 00:10:26.94 45
as needed the simulation environment allow us
-
00:10:26.94 - 00:10:28.33 15
to do into more
-
00:10:28.43 - 00:10:32.90 44
physics based more realistic modeling of the
-
00:10:32.90 - 00:10:33.60 7
sensors
-
00:10:33.60 - 00:10:37.21 38
One last point here in in creating the
-
00:10:37.21 - 00:10:37.77 13
simulation is
-
00:10:37.85 - 00:10:41.47 39
about driving scenarios and so what I'm
-
00:10:41.47 - 00:10:42.27 15
showing here is
-
00:10:42.35 - 00:10:45.96 39
something that is in the making because
-
00:10:45.96 - 00:10:46.53 4
it's
-
00:10:46.53 - 00:10:50.14 35
related to a standard that is being
-
00:10:50.14 - 00:10:51.02 20
constructing at ASAM
-
00:10:51.10 - 00:10:53.11 25
which is OpenSCENARIO 2.0
-
00:10:53.11 - 00:10:54.88 20
where you see a high
-
00:10:54.96 - 00:10:58.57 45
level language to create driving scenarios we
-
00:10:58.57 - 00:10:59.62 12
will have it
-
00:10:59.62 - 00:11:04.14 45
when the standard becomes available in little
-
00:11:04.14 - 00:11:05.14 6
bit of
-
00:11:05.24 - 00:11:05.65 4
time
-
00:11:05.65 - 00:11:10.00 38
Now let's turn to the software part so
-
00:11:10.00 - 00:11:10.38 5
again
-
00:11:10.48 - 00:11:14.83 43
we'll take our example and look at software
-
00:11:14.83 - 00:11:16.28 16
requirements and
-
00:11:16.38 - 00:11:20.73 43
how they are implemented and then simulated
-
00:11:20.73 - 00:11:21.60 10
So what we
-
00:11:21.60 - 00:11:25.91 45
have here is a software component that relies
-
00:11:25.91 - 00:11:26.77 8
on input
-
00:11:26.86 - 00:11:31.17 39
which is an occupancy grid where we see
-
00:11:31.17 - 00:11:32.03 14
occupied areas
-
00:11:32.13 - 00:11:36.43 31
and probability that this is an
-
00:11:36.43 - 00:11:36.91 10
object And
-
00:11:36.91 - 00:11:41.06 45
then we have a software requirement that says
-
00:11:41.06 - 00:11:41.98 9
that upon
-
00:11:42.07 - 00:11:46.22 43
detecting collisions threat we will produce
-
00:11:46.22 - 00:11:48.99 31
emergency breaking depending on
-
00:11:49.08 - 00:11:53.23 43
occupancy probability time to collision and
-
00:11:53.23 - 00:11:55.07 21
ego vehicle speed and
-
00:11:55.07 - 00:11:59.46 45
so in order to implement the software what we
-
00:11:59.46 - 00:12:00.24 7
propose
-
00:12:00.34 - 00:12:04.72 45
is the model based software notation and tool
-
00:12:04.72 - 00:12:05.60 8
that you
-
00:12:05.70 - 00:12:10.08 41
see here where in the very high level and
-
00:12:10.08 - 00:12:10.67 9
graphical
-
00:12:10.67 - 00:12:14.27 45
way you express the algorithms so here I will
-
00:12:14.27 - 00:12:14.59 3
not
-
00:12:14.67 - 00:12:18.27 41
go into the details but it is about going
-
00:12:18.27 - 00:12:18.59 7
through
-
00:12:18.67 - 00:12:22.28 45
iterations of these array cells and detecting
-
00:12:22.28 - 00:12:23.56 15
the objects and
-
00:12:23.56 - 00:12:27.08 43
understanding if breaking should be applied
-
00:12:27.08 - 00:12:28.41 18
so very high level
-
00:12:28.49 - 00:12:32.01 36
graphical notation that comes with a
-
00:12:32.01 - 00:12:33.50 27
certified code generator so
-
00:12:33.57 - 00:12:37.10 42
you never have to worry about the software
-
00:12:37.10 - 00:12:37.80 11
code itself
-
00:12:37.80 - 00:12:41.91 35
you only worry about the design and
-
00:12:41.91 - 00:12:43.28 24
understanding the design
-
00:12:43.37 - 00:12:47.49 37
is correct and meets the requirements
-
00:12:47.49 - 00:12:48.86 22
themselves and so once
-
00:12:48.95 - 00:12:50.05 12
this is done
-
00:12:50.05 - 00:12:54.36 44
You can go in simulation and so here we have
-
00:12:54.46 - 00:12:58.87 45
coupled the generated code so that we execute
-
00:12:58.87 - 00:12:59.85 9
the model
-
00:12:59.95 - 00:13:00.93 10
simulation
-
00:13:00.93 - 00:13:05.22 40
together with destination of the driving
-
00:13:05.22 - 00:13:06.65 19
scenario and we see
-
00:13:06.75 - 00:13:11.04 43
that here the car is going to break because
-
00:13:11.04 - 00:13:11.14 2
of
-
00:13:11.23 - 00:13:15.15 41
the pedestrian and then resume its course
-
00:13:15.15 - 00:13:18.73 44
Now one last topic I want to address is that
-
00:13:18.81 - 00:13:22.47 42
assessing you know the level of confidence
-
00:13:22.47 - 00:13:23.28 12
you can have
-
00:13:23.37 - 00:13:27.03 43
in in the application of emergency breaking
-
00:13:27.03 - 00:13:28.25 18
relies on coverage
-
00:13:28.25 - 00:13:31.83 44
of a large number of things first of all you
-
00:13:31.91 - 00:13:35.57 40
should cover all of the requirements the
-
00:13:35.57 - 00:13:36.71 18
scenarios and they
-
00:13:36.80 - 00:13:40.46 40
have to be tested second of all you must
-
00:13:40.46 - 00:13:40.95 11
ensure that
-
00:13:40.95 - 00:13:44.63 41
you cover all the situations that you are
-
00:13:44.63 - 00:13:45.04 8
going to
-
00:13:45.12 - 00:13:48.80 37
see in driving if not you have to add
-
00:13:48.80 - 00:13:48.96 9
scenarios
-
00:13:49.04 - 00:13:52.48 42
as we said and there is one last example I
-
00:13:52.48 - 00:13:56.21 45
wanted to illustrate here is that you have to
-
00:13:56.21 - 00:13:56.71 5
cover
-
00:13:56.79 - 00:14:00.52 38
through simulation all of the software
-
00:14:00.52 - 00:14:01.76 22
elements that there is
-
00:14:01.85 - 00:14:05.00 39
and here we see a tool which is a model
-
00:14:05.00 - 00:14:08.66 44
coverage tool for the software part and what
-
00:14:08.66 - 00:14:09.23 7
this is
-
00:14:09.31 - 00:14:12.98 43
telling us is that this condition has never
-
00:14:12.98 - 00:14:13.55 8
been set
-
00:14:13.63 - 00:14:17.30 44
to true in the overall executions of all the
-
00:14:17.30 - 00:14:20.47 43
simulations that we have performed and this
-
00:14:20.47 - 00:14:21.80 20
means that something
-
00:14:21.88 - 00:14:24.91 43
is missing and in fact the fact is that the
-
00:14:24.98 - 00:14:28.15 41
case that is missing is that when the two
-
00:14:28.15 - 00:14:28.50 8
vehicles
-
00:14:28.50 - 00:14:31.67 44
are going in the same directions and so this
-
00:14:31.67 - 00:14:31.81 2
is
-
00:14:31.88 - 00:14:35.06 44
an indication that there is a need either to
-
00:14:35.06 - 00:14:35.27 3
fix
-
00:14:35.34 - 00:14:38.51 44
the requirements or the model or to add test
-
00:14:38.51 - 00:14:38.86 5
cases
-
00:14:38.86 - 00:14:42.03 35
in order to be able to meet all the
-
00:14:42.03 - 00:14:42.24 12
expectations
-
00:14:42.31 - 00:14:45.49 45
in terms of coverage that was just an example
-
00:14:45.49 - 00:14:45.70 2
of
-
00:14:45.77 - 00:14:48.94 39
what kind of coverage but there is many
-
00:14:48.94 - 00:14:49.64 15
different kinds
-
00:14:49.64 - 00:14:52.95 43
of coverage that have to be that have to be
-
00:14:53.03 - 00:14:56.49 34
achieved so this is the end of the
-
00:14:56.49 - 00:14:56.95 16
presentation and
-
00:14:57.03 - 00:15:00.34 43
I will now turn it to a few questions thank
-
00:15:00.34 - 00:15:00.58 3
you