Autonomous driving is still a black box

There’s still a lot we don’t know about autonomous driving and it will be a real challenge to find the faults in artificial intelligence, says Timo Dolde, Senior Project Manager Data Analytics at P3 automotive GmbH. At this year’s CTI Automotive Diagnostics Conference he talked about the Systematical management of load collectives as success criteria. At the event we got him for an interview.

CTI: What are the biggest challenges in the automotive diagnostics?
Timo Dolde: I think the biggest challenges in the automotive diagnostics will be to find the faults in automotive driving cars because you know it’s artificial intelligence and it’s like black box so it’s not so easy to find a fault in deep rural networks.

CTI: And how does connectivity impact automotive diagnostics in your opinion?

Dolde: First of all I think it does impact the automotive diagnostics by the large amount of data which is coming because every car is sending the whole time.

CTI: Which approaches could be successful to deal with these big amount of data?

Dolde: I think the approach will be that something will be calculated in the car so that only differences betweeen algorithms will be sent to the back end so that the traffic, the data amount will be not so high.

CTI: You talked about the load collectives in your presentation, can you please more closely describe what it is about what is the usage and other information about it?

Dolde: So load collective data is data from the customer cars where you can see how does the customer use the cars, how often does the customer drive in which driving style, so high revs or low revs or something like that, and the usages that the OEM’s do know much more about the customers they do know exactly how they use the cars, what they need, what they want, which functions are used often, which are not so often used, so they can adapt their cars much better to what the customer needs.

CTI: Can you give me any example of any results of these data?

Dolde: Yes, one good example is in the route cause analysis, we did find some differences between cars with failure and cars without failure so that we were able to directly say okay this car did fail because of this kind of driving behaviour.

CTI: And you also talked about challenges of load collectives, can you tell me a little bit more about that?

Dolde: Yeah the biggest challenge is obviously the data quality so the processes has to be much more, much better in the companies because you know there are the different departments and they do not speak together so there should be central processes for the definition of load collectives.

CTI: What do you think about the automotive diagnostics conference here?
Dolde: I think it’s a high quality conference so what I heard was a lot of deep knowledge and all the important topics in diagnostics so I think the exchange of knowledge here is really good.

Timo Dolde is Senior Project Manager Data Analytics at P3 automotive GmbH.

Stay up to date