Senior Product Sales Manager and SME in ADAS, Autonomy and AI
Talk with discuss how Hybrid Virtual Digital Twin models are developed, validated, and used. The development and validation of high-fidelity physics-based digital twin sensor (radar and lidar) models will be used as examples.
The talk will discuss four key points in digital twin model development:
1. Accuracy – how closely the digital model adheres to actual physical behavior of the sensor
2. Flexibility – how well the digital model generalizes to multiple scenarios or use cases without overfitting to any single case
3. Adaptability – how well the digital model handles changing environmental conditions
4. Scalability – how rapidly the digital model can be built and deployed
I’ll also discuss the benefits of using hybrid analytics to combine real-world data and physics as follows:
1. Closely match simulation results with measurement data by calibrating model parameters
2. Increase the confidence in fit by using uncertainty quantification on parameters and outputs
3. Compensate for any unmodeled physics or other effects by modeling the difference between a physics model and data
Lastly, I’ll have some examples of use cases and potential time and cost savings that have been achieved using digital twins in simulation.