AURIX™ – TC3xx Artificial Intelligence
Artificial Inteligence is not limited to high end applications anymore and has entered the embedded world within the last decade.
When to use AI ?
AI as the bridge connecting humans with the digital world.
It aims to leverages human capacities for learning, perception and interaction. Therefore less information is lost during this “connection”.
Although the surface of AI remains rule-based reasoning, AI has deeper subfields going further than the standard programming way.
Machine Learning :
In this AI subfield, the computer is taught to do tasks with data without explicit programming. It is based on numerical and statistical approaches.
In brief, Machine Learning deals with the conception, the analysis, the optimization, the development and the implementation of such techniques.
For example, Neural Networks are part of the AI subdivision
Typicall Neural Network based application development steps:
- Create or find a model fitting the application needs
- Training Phase – Teach the network (on server or computing platform)
- Inference Phase – Deploy on target
- Validate the results
Deep learning is a subfield of Machine Learning with more that two Neural Network layers. Since it is computationally intensive, Deep Learning needs specialized HW.
Deep Learning based algorithms learn pattern from data by themselves, where human would even not have found out the relevant pattern in the data.
Where ML is only creating rules, DL is building more efficient decision rules by adding features extraction to the process.
AI in AURIX™-TC3xx
We enter a new era of innovation possibilities for the automotive industry, where AI brings much more than the well-known Automated Driving, with solutions regarding connected cars or safety concerns, among others.
- More and more in-vehicle sensors and ECUs lead to more and more data to compute.
- Embedded AI could reduce operational cost and improve vehicle safety.