国内工作半年了，先后对开源的仿真软件做了定制化实现，对高精地图、ros、场景定义等有了大量快速的接触。同时也搭建了团队的gitlab, jenkins服务, rosbag管理服务、webviz服务，pg地图服务等等，并且也投入了一些时间去搭建数据基础框架，但并没有实际业务需求。在传统oem，用DevOps去规范开发流程，以及探索数据业务、云服务等相对较新颖。相对，新兴的造车企业，会比较好的融合DevOps流程，开发上云。
本质上是开发ADS支撑工具及数据基础设施，但由于OEM内部的ads开发团队比较弱，这些支撑工具以及数据工具并不发挥价值。作为工具开发者，存在感会变弱。因为离核心业务较远，虽然技术很通用。
OEM’s role in emerging tech
OEM itself mostly doesn’t build tools itself, even like Ford, who hired a great team to develop CAE tools in early days and build them tremendous, still retired most of them or sell them as packages to pure software vendors later.
on the other side, the pioneers or leaders of the industry, usually build and apply new methodlogy before the market is there. and after the market is there, they can have the professional team to take in charge. the leaders take the role as incubators, as many Internet startups got the ideas from giants e.g. Google, Facebook e.t.c
ADS at this moment is the emerging tech, there is no mature solution or tool chain for ADS, so the leaders of automobile makers and suppliers are the guys who should take the lead to set up standards and tools till ADS market goes to mature. and the giants in automobile should catch this great opportunity to take control in long term.
the problems of Chinese OEMs (sadly there is no strong Chinese suppliers yet), they are customed to be followers, when US or German auto makers do the research and make the new bussiness stratagey, Chinese OEMs will follow. as the blog said, Chinese has no friend, neither Chinese goverment. be the followers or imitators is minimal loss strategy for most Chinese companies as well. and this is also why west coutry companies are afriad of the competition from Chinese. Chinese take copy others ideas as a strategy to success, rather than a shame.
a tool developer
机缘巧合，职业最初，就属于工具支撑组，开发和维护工具。随着，ads的爆发，more and more software skills are required by tradititional automobile makers and suppliers; and they prefer transfer themselves to modern digital driven. but a special tool is never the core product of auto makers.
there comes a time, the OEM engineers asked, why you guys build tools, rather than buy from suppliers. The right thing for OEMs is to integrate components, rather than building tools.
the cost of building tools by OEM itself is really too much after a while we realized. But at first, OEMs thought it was too costy to buy commecial tools, which gives survival space of a small team to grow inside OEM teams.
actually at the early time, small team dreams big, especially in these days, many open source projects can be used freely. the small team almost support every aspect related to software tools, and as well to add features to simulation platform, which is also based on an open source project lg-sim, software manager looks like a good role in the team. but the reality is, to customize any tool, will cost a huge energy for a small team, as we are not involved in these open source projects. more, the manager team doesn’t consider an independent tool team. and even like this, we are moving slowly.
infrasturcture for ABC
for ADS startups, they emphasize their infrasturcture features with :
- cloud platform
- remote monitor/control system (for vehicle team management)
- OTA
- data infrastructure(storage, management, analysis)
and in single vehicle features with powerful computing ability, from GTC 2018 speak, most startup use the powerful Xdriver GPU；
- WeRide
- Momenta
- Tusimple
- AutoX
- Roadstar.AI (dead)
- plusAI
- ponyAI
all these looks very Internet, but too far from massive vehicle product. the other mind is these teches are actually common for average engineers, but the chanllenge is to build a team, to realy make it work.
survive in OEM
it’s maybe more interested to take myself as an ABC(AI, big data, cloud computing) guy to find application scenarios in automobile industry product development. while ABC application scenarios are not matured yet of course, usually it is driven by leader teams, when it comes to drive by market, may be too late.
beyond automotive, industry IoT maybe an even general domain to quickly apply ABCs. there should be three steps:
familiar with the tech methods
understand the hurting points of the industry
product and management update -> value return
the life leap happens to a few lucky guys. most guys are common and grow in a linear speed. there is a burden the more you know, the harder your life. sadly …
digitalize
OEMs are involved more and more in transfering digital, smart. previously most invovled in market, ads strategy, supply chain management; and as connected vehicles, self driving, future mobility service bursting in recent years, digitalization is sinking in vehicle product R&D. AI, big data, cloud, blockchain e.t.c. new techs will be transfering the traditional industry. hopefully we can define the new bussniess together.