Talk:VaMP

Influence[edit]

The VaMP is probably the most influential driverless car ever. I started this article using some of the material in Ernst Dickmanns and the EUREKA Prometheus Project. Is anybody out there in a legal position to upload images of the VaMP? Science History 11:29, 15 June 2007 (UTC)[reply]

It's funny how the 2007 Urban Grand Challenge is trying to repeat something the VaMP was able to do 12 years ago. And the VaMP was even faster. The field has become a bit stagnant in the past decade. Willingandable 16:33, 21 August 2007 (UTC)[reply]

What happened at the end of the project?[edit]

It is interesting to me that there is no work mentioned since 1995 (12 years before this writing). What happened? My instinct is that there was some kind of setback (financial? technical? personal? who pulled the plug? and why?) This should be mentioned, for the sake of NPOV, and also because no story is complete without the ending. Does anyone know the facts? Could we contact someone who worked on the project towards the end? -- CharlesGillingham 22:01, 23 August 2007 (UTC)[reply]

I think Ernst Dickmanns retired, and many of his team took jobs in the car industry (mostly Audi and Daimler-Benz?), where they don't publish their recent results. But both Audi and Daimler-Benz seem to have a small fleet of driverless research vehicles. Willingandable 16:21, 24 August 2007 (UTC)[reply]

One possibility is that Dickmann's approach suffered critiques that are not documented in this article. There is a hint of critique in this source where Dickmann writes

The argument sometimes heard that these systems will be 'closed' as opposed to neural-net-based ones is not intelligible from this point of view

This suggests that Dickmann's approach may have been superceded by approaches based on neural nets. Another possibility is that limited computing power ultimately put a limit on how far the VaMP could be improved. Dickmann hints at this when he writes (in the same source):

Computing power is arriving now for handling real word problems in real-time. Lack of robustness encountered up to now due to black-and-white as-well-as edge-based image understanding can now be complemented by area-based representations including color and texture, both very demanding with respect to processing power.

And finally (most interestingly) he hints that his approach is being abandoned by many researchers:

Taking advantage of well suited methods in competing approaches and combining the best of every field in a unified overall approach will be the most promising way to go. The good old stuff should not be discarded too early

(Italics mine).

I would not dream of adding to the article based on this paltry source, since that would be original research. ---- CharlesGillingham 21:45, 25 September 2007 (UTC)[reply]

References Needed[edit]

I'm considering marking this article as unreferenced. I have no doubt the information here is correct, but I think the original author has sources that would be useful to further research on the subject. Please provide them! ---- CharlesGillingham 21:45, 25 September 2007 (UTC)[reply]

Vision vs GPS[edit]

I suspect that the use of vision only vs vision plus GPS is based on the VaMP project focusing on collision avoidance only. Was there any attempt to autonomously navigate a set course or to autonomously navigate from one point to another? Some clarification seems necessary regarding that difference. Moretz (talk) 10:51, 12 April 2010 (UTC)[reply]