Software development challenges in automotive projects
summary
Automotive projects must meet ISO26262 and ASPICE while handling practically infinite driving scenarios, and regulators still expect 100% verification even though automated driving breaks that assumption. Radar processing converts electromagnetic waves to a radar data cube → point cloud → environment model for object detection, enabling hidden object and vulnerable road user detection. Workflows rely on CI/CD with Google Test, Doxygen, PlantUML and qualified tools plus heavy automation to get fast feedback. EU road deaths remain above 20,000 per year, so robustness not just performance is the primary metric.
AI-written summary · about
chapters (11)
- 0:00Automotive software challenges
- 3:13Why development is complex
- 7:58Safety standards impact
- 13:20Automotive radar complexity
- 17:38AI performance vs understandability
- 20:08Agile in automotive challenges
- 24:02Optimizing processes and tools
- 28:56Automate development tasks
- 33:59CI/CD in software delivery
- 35:02Key takeaways
- 36:10Q&A on embedded languages and Tesla Vision
related talks
47mEmbedded Systems Architecture
20mDriving Intelligence: The Software Shift Shaping Tomorrow’s Cars
19mDriving Automotive with Open Source: From Compliance to Collaboration
44mHandling Complexity in Software Development Projects
57mFinding, Managing, Preventing Vulnerabilities: An Automotive Perspective
14mPANORover: Autonomous Driving System Development Platform
53mWhat’s the Reality of RISC-V in Automotive?
20mThe Challenges in Supporting Multiple Safety Levels in the Same System