Live! From Embedded World, ISSCC, and (sorta) Mobile World Congress

Engineering & simulationCartesium Studio

The tool allows embedded developers to create and link a machine learning library directly into their program, enabling machine learning training and inference to run locally on the microcontroller without needing cloud data scientists.

“we are announcing today a studio, which is a development environment that we let any embedded developers create a machine learning library inside the studio, and then they can link the library into their program, and they get the machine learning inside the microcontroller.”

Engineering & simulationTI Jacinto 7 processor

The processor runs multiple deep learning neural networks concurrently for automotive applications, specifically identifying vehicle classes and detecting parking spots for automated valet parking.

“One is a 68 deep convolutional general network. The other is 48 deep network. All 3 running at the same time, in the DSP, and rendered. And together, so essentially 2 different applications running concurrently in the same SOC.”