Infineon Technologies is to aiming the vector processing unit in the coming generation of Aurix automotive microcontroller at AI applications for the battery management system.
The company says it has the largest global market share for automotive microcontrollers, so the use of the vector unit in the latest TC4x family for AI with ASIL D functional safety is a key move forward.
Aurix is based on the 32bit TriCore architecture with its own instruction set architecture and the TC4x family includes a parallel processing unit with a vector-based single instruction multiple data (SIMD) for AI acceleration alongside up to to six TriCore v1.8 cores running in lockstep at up to 500 MHz.
“In the next generation of Aurix we will include a vector processing unit for a parallel processing unit to bring AI directly into the car, for example for the battery management system.” said Christian Feldmann, vice president of technical marketing for automotive microcontrollers at Infineon.
The TC4x family was launched in November 2024 and is set for mass production in 2025 for customers including Continental and Marelli.
Feldmann points to the use of machine learning and AI for the vehicle battery management system (BMS).
“A battery pack can have up to 10,000 cells but manufacturers monitor only a few cells. To offset the inaccuracy of this the manufacturers include a buffer which uses energy and reduces range. With Aurix we will be able to determine the charging state more accurately and increase range up to 20% so smaller batteries will be sufficient, driving down prices and making EVs more affordable.”
“The safety condition of the battery is the second most important factor and we can use AI to measure the state of health (SoH) more precisely and extend the lifetime of the battery by 30% and reduce fast charging times by 20%,” he said.
Infineon is also using AI to automate its chip layouts, achieving layouts in 12 days rather than 60, and using AI for code development.
Leave a comment