Examples of projects involving data-based services
Automated labelling of image data
A neural network had to be trained to assess image data for an autonomous driving project in the automotive industry with the aim of improving the environment detection process using cameras. Machine-readable descriptions needed to be assigned to more than 50 million images of road scenes using an automated process. We are currently developing high-performance algorithms for this purpose, together with a toolkit for the automated labelling of image data.
Identifying component faults by analysing vibrations
A defective vehicle component developed reproducible cracks while the vehicle was being driven, which had no obvious cause. We carried out an analysis which involved correlating the component’s vibration data with the lateral and longitudinal acceleration of the vehicle. This allowed us to identify that the thickness of the material used to make the component was incorrect and was causing the problem, because it had led to destructive vibrations.
Evaluating the quality of the road surface using vehicle shock absorbers
The aim of the project was to evaluate the quality of the road surface on the basis of the behaviour of vehicle shock absorbers. An external service provider specialising in data analysis was unable to produce any plausible results despite the use of machine learning and neural networks. The company had no knowledge of the industry and no understanding of driving dynamics. The data analysts from Bertrandt succeeded in drawing valuable conclusions about the quality of the road surface on the basis of their industry expertise, mathematical-statistical methods and application-specific algorithms.