Speaker: Hualiang Shi, Level 5 Self-Driving Division, Lyft Corp.

Meeting Date: Thursday, August 12, 2021
Time: Checkin via WebEx at 11:50 AM; Presentation at 12:00 noon (PDT)
Cost: none
Reservations: eps2108a.eventbrite.com
Summary: Robo-taxi has various potential applications, including the reduction of risky driver behaviors, transportation cost reduction, and lowered carbon emission, and the enhancement of road safety, independence for seniors and people with disabilities, and human productivity. It’s full of opportunities and challenges. In this presentation, we will share our qualification methodology as well as some technical lessons. First, the use cases for robo-taxi and conventional cars are compared. Due to the differences of use cases, the existing industry standards might not be suitable for the qualification of robo-taxi. Second, the customizations of stress profiles by instrumenting vehicles with accelerometers and thermocouples, and leveraging weather station data on the basis of operational design domain are demonstrated. This kind of stress profile customization can balance the design efforts and time-to-market, and avoid under-design or over-design. Vibration stress profile, temperature cycle stress profile, temperature humidity stress profile, and UV exposure stress profile are used as case studies. Vibration test and shock test of various PCB boards are used to show the impact of stress profiles. Third, several hardware failures and issues are discussed, including air bubbles inside liquid coolant, down-selection of camera connector sealing material, the buckling of cold plates, the leakage of liquid coolant, the electrical-open failure of LEDs, fatigue cracking of a Lidar bracket, the electrical-open failure of a mechanical relay, blistering of exterior paint, corrosion on a PCB board, the electrical breakdown of a power supplier unit, and the malfunction of Radar. Different approaches are applied to understand these failures and issues, including analytical modeling, numerical regression analysis, Finite Element Analysis, FTIR, I-V curve, SEM/EDX, 3D X-ray, and 3D depth profiling. Fourth, six methods to mitigate the low sample size issue are shared, including prioritization of tests by DFMEA, sequential waterfall tests, combination of tests at different levels, combination of tests at different phases, stress-to-fail, and some statistical methods.

Bio: Dr. Hualiang Shi is a staff reliability engineer and engineering manager at Lyft’s Level 5 self-driving division. Before Lyft, Dr. Shi spent more than fifteen years in the field of semiconductor and consumer electronics. His job roles included Integration Engineer at Apple’s Silicon Engineering Group, Senior Reliability Engineer at Apple’s Product Integrity Group, Staff Quality and Reliability Engineer at Intel’s Corporate Quality Network, Senior Process Engineer at Intel’s Silicon Integration and 3D Package Integration Group, Process Engineer at Intel’s Epoxy Module group, and an Internship at Tokyo Electron Limited. Dr. Shi got his Ph.D degree in Physics from Dr. Paul S. Ho’s group at the University of Texas at Austin at 2010 and B.S degree in Physics from the University of Science and Technology of China in 2003. Dr. Shi was on the technical committee for 3D/Packaging for the 2016 IEEE International Reliability Physics Symposium (IRPS) and for Packaging and Assembly Level FA at the 43rd International Symposium for Testing and Failure Analysis (2017). He served as a journal reviewer for fourteen journals, and published four book chapters and eighteen papers, and holds five patents in the fields of semiconductor backend-of-the-line (BEOL), semiconductor IC packaging, and autonomous vehicles.