HomeIHelpISite mapISearch
ModelsCompanyNews & EventsExperienceContact Us
Stanley.

Stanley.

Innovation
Suspension
Body
Assistance systems
Infotainment
Future mobility
New Small Family
Passat Junior
Stanley
Drive
Stanley  |  Technology  |  Computing strategy
Computing strategy – independent thinking for better decisions.

Computing strategy – independent thinking for better decisions.

+
The program for controlling “Stanley” was not in a single block, rather it was broken down into several individual modules, each with their own set of tasks. In all, 31 software modules with a total of around 100,000 lines of source text (programming lines) ensured that “Stanley” made steady progress. By way of comparison: the Windows XP operating system has over 40 million lines of source text.
The program for controlling “Stanley” was not in a single block, rather it was broken down into several individual modules, each with their own set of tasks. In all, 31 software modules with a total of around 100,000 lines of source text (programming lines) ensured that “Stanley” made steady progress. By way of comparison: the Windows XP operating system has over 40 million lines of source text.
The computing strategy for “Stanley” can be described in the following simplified form: to enable “Stanley” to find its way, the Advanced GPS supplied the software with the current position, while the five laser sensors were used to scan the terrain ahead up to just in front of the vehicle for obstacles. The maximum possible speed was determined based on the information received from the camera and the long-range radar. Here, the software did not rely on the use of stored topographical data for the race area, although this was allowed; instead, it processed the information fed from the sensors during the journey in real time.
+
The computing strategy for “Stanley” can be described in the following simplified form: to enable “Stanley” to find its way, the Advanced GPS supplied the software with the current position, while the five laser sensors were used to scan the terrain ahead up to just in front of the vehicle for obstacles. The maximum possible speed was determined based on the information received from the camera and the long-range radar. Here, the software did not rely on the use of stored topographical data for the race area, although this was allowed; instead, it processed the information fed from the sensors during the journey in real time.
+
Programming the detection and evaluation of obstacles proved to be exceedingly difficult. How should the system react to stones or bushes that were in the way? Whereas bushes could be driven over without difficulty, boulders would soon bring the vehicle’s race to a premature end. The reaction to moving obstacles – other race contenders, for instance – also had to be taken into account. Where and when was it possible to overtake? For this, the software had to undergo rigorous “training”.
Programming the detection and evaluation of obstacles proved to be exceedingly difficult. How should the system react to stones or bushes that were in the way? Whereas bushes could be driven over without difficulty, boulders would soon bring the vehicle’s race to a premature end. The reaction to moving obstacles – other race contenders, for instance – also had to be taken into account. Where and when was it possible to overtake? For this, the software had to undergo rigorous “training”.
Interpretation of the laser data was greatly improved, for example, using an automatic learning program. As a result, the error rate for interpretation of the information dropped from 12 percent to just 0.02 percent. Numerous exhaustive test drives were furthermore conducted to “teach” the system how to devise effective solution strategies by following the example of a human driver. This gave “Stanley” the degree of independence and rapid reactions which would ultimately produce the crucial extra pace that secured the autonomous Touareg victory.
+
Interpretation of the laser data was greatly improved, for example, using an automatic learning program. As a result, the error rate for interpretation of the information dropped from 12 percent to just 0.02 percent. Numerous exhaustive test drives were furthermore conducted to “teach” the system how to devise effective solution strategies by following the example of a human driver. This gave “Stanley” the degree of independence and rapid reactions which would ultimately produce the crucial extra pace that secured the autonomous Touareg victory.
VOLKSWAGEN AGIVolkswagen international
© Volkswagen 2011ILegal