Prometheus (research program)

from Wikipedia, the free encyclopedia

The EUREKA -PROMETHEUS project ( PRO gra M me for a E uropean T raffic of H ighest e fficiency and U nprecedented S afety , 1986-1994) came at the initiative of Ferdinand Panik (then Daimler-Benz) and his colleagues from the Group research by most of the other European vehicle manufacturers. In this largest European project to date to improve the efficiency, environmental compatibility and safety of road traffic with a funding volume of over 700 million ECU , almost all major European suppliers and other electronics companies that were not previously active in the automotive sector, as well as one Numerous scientific institutes.

content

The fifteen-month definition phase (10/1986 - 12/1987) was followed by the actual processing (1/1988 - 12/1994). The project structure provided for seven sub-projects. There were four sub-projects for the basic research carried out by scientific institutes:

  • PRO ART
  • PRO-CHIP
  • PRO-COM
  • PRO-GEN.

From the three industrial research sub-projects

  • PRO-CAR
  • PRO-NET
  • PRO-ROAD

Common European Demonstrators (CED) were derived, in which thematically related functions were presented by mostly several vehicle manufacturers in their test vehicles. These were presented to a wider public in board member meetings (01/89 Munich, 09/91 Turin, 10/94 Paris). In detail it was

  • CED 1: Vision Enhancement,
  • CED 2: Safe vehicle operation (Proper Vehicle Operation), again divided
  • * CED 2.1: Monitoring of driving stability (Friction Monitoring and Vehicle Dynamics)
  • * CED 2.2: assistance in tracking ( Lane Keeping Support )
  • * CED 2.3: Visibility Range Monitoring
  • * CED 2.4: Driver Status Monitoring
  • CED 3: collision avoidance (Collision Avoidance)
  • CED 4: Co-operative Driving
  • CED 5: Autonomous, intelligent cruise control (Autonomous Intelligent Cruise Control)
  • CED 6: Automatic Emergency Call
  • CED 7: Fleet Management
  • CED 9: Dual Mode Route Guidance
  • CED 10: Travel and Traffic Information Systems.

The originally planned demonstrator CED 8 (test fields) turned out to be incompatible with the funding structure and the schedule and was therefore dropped. A total of 27 individual functions were hidden behind these demonstrators, 13 of which were later found in series products, especially in driver assistance systems. Another 6 functions were later implemented on the basis of machine vision. Even today, most driver assistance systems can be referenced to PROMETHEUS preparatory work, which is why PROMETHEUS is considered a showcase project for particularly successful pre-competitive cooperation in a European framework.

The fact that PROMETHEUS is often mentioned in connection with automated driving is less due to the official objectives of this program, in which automated driving did not occur at all, e.g. In some cases it was even explicitly rejected, but rather because of the results shown in the context of PRO-ART on CED points 2.2 (Lane Keeping Support), 3 (Collision Avoidance) and 5 (Autonomous Intelligent Cruise Control). The higher management levels in the automotive industry were initially skeptical to negative about machine vision for driving vehicles.

PRO ART

The sub-project benefited from the ten years of preparatory work that the group had carried out at the University of the Bundeswehr in Munich (UniBwM) led by Prof. Ernst D. Dickmanns . In addition to the visual control of comparatively simple real processes through real-time image sequence processing (rod balance on an electric cart and docking of a sluggish hovercraft as a satellite simulator), a Hardware in the Loop (HiL) simulation circuit was set up there, which enables the investigation of the autonomous visual guidance of land and Aircraft with real sensors and specific computers in a circle. Due to the positive results, a 5-ton Mercedes-Benz 508 D panel van was purchased in 1985 to convert it into a “test vehicle for autonomous mobility and computer vision” (VaMoRs) with its own 220 V power generator and the required sensors and actuators. From 1986 until the new century, VaMoRs became one of the most successful test vehicles for autonomous visual vehicle guidance worldwide.

Work done in advance of PROMETHEUS

In German industry, only Volkswagen had attempted visual road vehicle guidance. At the UniBwM, the topic was dealt with increasingly broadly from 1982 onwards through third-party research contracts; In 1986 the “Chances and Challenges” of this new technology approach were presented and discussed for the first time at the ICTS Symposium on Human Factors Technology for Next-Generation Transportation Vehicles in Amalfi.

Definition phase of PRO-ART as part of PROMETHEUS

At this point in time, contacts between Daimler-Benz AG (DBAG) and UniBwM had already been established in order to further develop this technology approach in joint projects with public funding. For approval by the DBAG Research Board in December 1986, the capabilities of VaMoRs for autonomous lateral and longitudinal guidance were demonstrated by means of image sequence analysis on a slide plate. On this basis, it was decided that during the definition phase for the research project PROMETHEUS provided automatic transverse guidance by inductive one ', through the flexible computer vision' to replace electromagnetic fields after the presentation, state-of-the-art reviews for machine vision. Instead of the costs for the cables to be installed in the middle of the lane, digital technology and software development were financed. Around a dozen European companies in the automotive industry and around 60 research and university institutes from European countries took part in this project.
Many alternative approaches have been investigated both for sensors ( radar and laser range finders of various classes) and for software ( neural networks , classic AI approaches, engineering methods and their combinations). A project part 'PRO-ART' (from ARTificial Intelligence) was discussed for research in this area. The final report of the UniBwM on the Pro-Art definition phase with a time and milestone plan (others were submitted by the other European research groups involved) led to new approaches in this area. In the summer of 1987, the UniBwM research group was able to demonstrate that the VaMoRs research vehicle can recognize the road and its curvature. In addition, at adapted speeds of up to 96 km / h (maximum speed of VaMoRs), more than 20 km were covered on an obstacle-free track, completely autonomously, lengthways and crossways.

Work in the PROMETHEUS program

The first goal of the cooperation between DBAG / UniBwM in Pro-ART was defined to duplicate the second-generation image sequence preprocessing system (BVV2) developed in-house in an expanded form. For an interim demonstration on the occasion of the PROMETHEUS Board Member Meeting in 1991 in Turin, the test vehicle 'Vision Technology Application' (VITA, later VITA I), a 7-ton panel van from DBAG, was upgraded. The software for visual perception was to be expanded to such an extent that, in addition to lane and lane recognition, another vehicle could be discovered and tracked on the road. The distance to this vehicle should be continuously determined purely visually (no radar or lidar ). Building on this, the ability to drive autonomously in convoy at a safe, speed-dependent distance to a standstill was developed. The success also led to a better acceptance of the intention to develop a sense of sight for road vehicles among the upper echelons of the auto industry. For the PROMETHEUS final demonstration in Paris in 1994, a much more ambitious goal was then defined: two vehicles of the Mercedes-Benz S-Class ( W 140 ) should be able to drive fully autonomously with only visual perception in normal three-lane autoroute traffic. They were expanded with appropriate sensors and actuators as well as the necessary computer systems. DBAG named their vehicle VITA II (as Prometheus demonstrator 'CED 302'), UniBwM named their VaMoRs-PKW or VaMP for short ('CED 303'). DBAG was responsible for the complex mechanical modifications and extensions of both vehicles, while the 'vehicle eye' and the new transputer system for image sequence analysis as well as the entire software for visual perception and behavior control were the responsibility of UniBwM.

VaMP (test vehicle for autonomous mobility and computer vision)

The following skills could be demonstrated in October 1994 in public three-lane motorway traffic at Charles-De-Gaulles Airport with guests on board:

  • Free lane driving up to the maximum permitted speed in France of 130 km / h,
  • Transition to convoy driving behind any vehicle in front with a speed-dependent distance;
  • Detection and tracking of up to six vehicles with bifocal vision each in the front and rear hemisphere in the own lane and the directly adjacent adjacent lanes.
  • Independent decision to change lanes and to carry out it autonomously after the safety driver had given the clearance by setting the blinkers.

A total of more than 1000 km were driven accident-free on the three-lane autoroutes around Paris. Technical details on the methods and hardware systems can be found in various publications. When looking back, it must be taken into account that the computing power of microprocessors available at the time was around a factor of 100,000 less than in 2014. The costs of such a system were correspondingly high. This was also one of the reasons why the industry decided to use radar and lidar as sensors in the next development step towards standard collision avoidance / distance maintenance (CED 3).

The two CED 3 vehicles mentioned were the only ones in the PROMETHEUS project (even worldwide) that were able to demonstrate the above-mentioned services under good environmental conditions. For almost all the other European participants in the final demonstration, image sequence analysis only served special individual tasks such as lane, lane, individual object and sign recognition. The immediately following 'International Symposium on Intelligent Vehicles' allowed an assessment of the international status of development: Through the PROMETHEUS project, and especially through Pro-ART, Europe had achieved a leading position in the field of seeing road vehicles. These findings formed the basis for the later defined objectives for self-driving vehicles .

PRO-CHIP

PRO-CHIP was a sub-project in the field of basic research, mostly carried out by scientific institutes. The objective of PRO-CHIP was to research hardware that is adapted to the challenges of intelligent signal processing in the vehicle. The focus was on sensor technology, among other things, image converters with high dynamics, and computer technology, among other things, with parallel computers for image processing.

PRO-COM

PRO-COM was a sub-project in the field of basic research, mostly carried out by scientific institutes. The objective of PRO-COM was to develop methods and standards for data communication that would meet the requirements of vehicle-vehicle and vehicle-infrastructure communication.

PRO-GEN

PRO-GEN (alternatively Pro-General) was a sub-project in the field of basic research, mostly carried out by scientific institutes. The objective of PRO-GEN was to develop traffic scenarios for the assessment and possible implementation strategies. For the evaluation, the PROMETHEUS goals of traffic efficiency and road safety were considered in particular, but also questions of acceptance of the then new technology concepts.

PRO-CAR

PRO-CAR was a sub-project from the field of industrial research. In contrast to the sister projects PRO-NET and PRO-ROAD, the focus is on vehicle-independent solutions to improve road safety. As part of this sub-project, the Common European Demonstrator CED 1: Vision Enhancement, CED 2: Safe Vehicle Operation (Proper Vehicle Operation), in turn divided into CED 2.1: Monitoring of driving stability (Friction Monitoring and Vehicle Dynamics), CED 2.2: Support in lane keeping (Lane Keeping Support), CED 2.3: Visibility Range Monitoring and CED 2.4: Monitoring the driver status (Driver Status Monitoring), CED 3: Collision Avoidance and CED 5: Autonomous, intelligent speed and distance control (Autonomous Intelligent Cruise Control). Many of the functions presented in these projects were later implemented in series in driver assistance systems, such as adaptive cruise control , lane keeping support , night vision systems and drowsiness warnings .

PRO-NET

PRO-NET was a sub-project from the field of industrial research with a focus on vehicle-to-vehicle communication and the potential improvement in traffic performance and safety. However, the communication and localization techniques available for the desired functionality were not sufficient, so that only a few functions were actually represented in the Common European Demonstrator 4: Co-operative Driving. It was only around 20 years later that the technology was ready for some of these ideas to be shown in research projects such as Ko-FAS and in a field test simTD (2008-2013). Whether and when the ambitious functions of cooperative driving will be used in series production remains unclear.

PRO-ROAD

PRO-ROAD was a sub-project from the field of industrial research. In this sub-project in particular, the PROMETHEUS vision of a holistic solution to traffic problems emerged. Some of the hopes could later be fulfilled, such as dynamic route guidance, fleet management and intermodal travel information. The automatic emergency call is also about to be introduced. Other PRO-ROAD functions such as the infrastructure-based accident warning COMPANION do not seem to have any implementation perspective to this day, as the advance payment for the infrastructure equipment and the operating costs are proving to be an obstacle to implementation.

Web links

Individual evidence

  1. ↑ idw : FHTE awards honorary professor title to Dr.-Ing. Ferdinand panic
  2. ^ H. Zimmer: PROMETHEUS - A European research program for the design of future road traffic. In: Research Society for Roads and Transportation: Road Traffic Technology. Volume 34/1, 1990.
  3. ↑ HH Braess, G. Reichart: Prometheus: Vision of the “intelligent automobile” on the “intelligent road”? Attempt at a critical appraisal. Part 1. In: ATZ Automobiltechnische Zeitschrift. 4/1995, pp. 200-205.
  4. ↑ a b c d e H. Winner, M. Graupner: PROMETHEUS - Which visions became reality? In: Proceedings 17th VDA Technical Congress, Filderstadt, March 19 and 20, 2015. pp. 25–48.
  5. ↑ a b c d e f H. H. Braess, G. Reichart: Prometheus: Vision of the “intelligent automobile” on the “intelligent road”? Attempt at a critical appraisal. Part 2. In: ATZ Automobiltechnische Zeitschrift. 6/1995, pp. 330-343.
  6. ^ HG Meissner: Control of dynamic systems based on pictorial information. Dissertation. UniBwM, LRT; July 20, 1982.
  7. ^ ED Dickmanns, A. Zapp: Guiding Land Vehicles Along Roadways by Computer Vision. In: Proc. Congres Automatique, AFCET, Toulouse, 1985. pp. 233-244.
  8. ^ W. Zimdahl, I. Rackow, T. Wilm: OPTOPILOT - a research approach for lane recognition and lane guidance in road vehicles. In: VDI reports. No. 162, 1986, pp. 49-60.
  9. ^ ED Dickmanns, A. Zapp: A Curvature-based Scheme for Improving Road Vehicle Guidance by Computer Vision. In: Mobile Robots. SPIE Proc. Vol. 727, Cambridge, Mass., Oct 1986, pp. 161-168.
  10. ^ ED Dickmanns: Computer Vision in Road Vehicles - Chances and Problems. ICTS-Symposium on Human Factors Technology for Next-Generation Transportation Vehicles. Amalfi, Italy, 16.-20. June, 1986.
  11. ^ ED Dickmanns: 4-D Dynamic Scene Analysis with Integral Spatio-Temporal Models. 4th Int. Symposium on Robotics Research, Santa Cruz, 1987. In: RC Bolles, B. Roth: Robotics Research. MIT Press, Cambridge 1988, pp. 311-318.
  12. ^ ED Dickmanns: Dynamic Vision for Perception and Control of Motion. Springer-Verlag, 2007, p. 214.
  13. ^ ED Dickmanns: PROMETHEUS 11 170 Integrated Approaches: State of the art review. March 1987, pp. 1-35.
  14. ^ ED Dickmanns, V. Graefe, W. Niegel: Final report definition phase PROMETHEUS. Pro-Art of the UniBw Munich, Nov. 1987, pp. 1-12.
  15. ^ A. Zapp: Automatic road vehicle guidance by computer vision. Dissertation UniBwM, LRT, September 8, 1988.
  16. ^ A b E. D. Dickmanns, V. Graefe: a) Dynamic monocular machine vision. Machine Vision and Applications. Springer International, Vol. 1, 1988, pp. 223-240. b) Applications of dynamic monocular machine vision. (ibid), 1988, pp. 241-261.
  17. ^ ED Dickmanns: Dynamic Vision for Perception and Control of Motion. Springer-Verlag, 2007, p. 216.
  18. ^ A b E. D. Dickmanns, T. Christians: Relative 3-D-State Estimation for Autonomous Visual Guidance of Road Vehicles. In: Int. Conf. on Robotics and Autonomous Systems. Vol. 7, Elsevier Science Publ. 1991, pp. 113-123.
  19. ↑ ED Dick Mann, B. Mysliwetz: Recursive 3D Road and Relative ego-state recognition. In: IEEE Transactions PAMI. Vol. 14, No. 2, Special Issue on 'Interpretation of 3-D Scenes', Feb 1992, pp. 199-213.
  20. ↑ B. Mysliwetz: Parallel computer-based image sequence interpretation for autonomous vehicle guidance . Dissertation. UniBwM, LRT; August 10, 1990.
  21. ↑ C. Brüdigam: intelligent driving maneuvers of seeing autonomous vehicles in a motorway-like environment. Dissertation. UniBwM, LRT; June 22, 1994.
  22. ↑ R. Behringer: Visual recognition and interpretation of the course of the lane by computer vision for an autonomous road vehicle. Dissertation UniBwM, LRT; March 14, 1996.
  23. ^ ED Dickmanns, R. Behringer, D. Dickmanns, T. Hildebrandt, M. Maurer, F. Thomanek, J. Schiehlen: The Seeing Passenger Car 'VaMoRs-P'. In: I. Masaki (ed.): Proc. of Int. Symp. On Intelligent Vehicles '94, Paris. 1994, ISBN 0-7803-2135-9 , pp. 68-73.
  24. ^ A b E. D. Dickmanns: Dynamic Vision for Perception and Control of Motion. Springer-Verlag, 2007, Fig.11.29
  25. ^ F. Thomanek, ED Dickmanns, D. Dickmanns: Multiple Object Recognition and Scene Interpretation for Autonomous Road Vehicle Guidance. In: I. Masaki (ed.): Proc. of Int. Symp. On Intelligent Vehicles '94, Paris, Oct. 1994. pp. 231-236.
  26. ↑ F. Thomanek: Visual recognition and condition estimation of several road vehicles for autonomous vehicle guidance. Dissertation UniBwM, LRT; January 25, 1996.
  27. ^ ED Dickmanns: Dynamic Vision for Perception and Control of Motion. Springer-Verlag, 2007, Fig.11.22
  28. ↑ J. Schiehlen: Camera platforms for actively seeing vehicles. Dissertation UniBwM, LRT; June 2, 1995.
  29. ↑ M. Schmid: 3-D recognition of vehicles in real time from monocular image sequences. Dissertation. UniBwM, LRT; September 12, 1994.
  30. ↑ B. Ulmer: VITA II - Active Collision Avoidance in Real Traffic. In: I. Masaki (ed.): Proc. of Int. Symp. On Intelligent Vehicles '94, Paris, Oct. 1994, IEEE. ISBN 0-7803-2135-9 .
  31. ↑ Von Holt: Tracking and Classification of Overtaking Vehicles on Autobahns. In: I. Masaki (ed.): Proc. of Int. Symp. On Intelligent Vehicles '94, Paris, Oct. 1994. ISBN 0-7803-2135-9 , pp. 314-319.
  32. ↑ J. Schiehlen, ED Dick's: A Camera Platform for Intelligent Vehicles. In: I. Masaki (ed.): Proc. of Int. Symp. On Intelligent Vehicles '94, Paris, Oct. 1994. IEEE, ISBN 0-7803-2135-9 , pp. 393-398.
  33. ↑ Masaki (Ed.): Proceedings of International Symposium on Intelligent Vehicles '94, Paris, Oct. 1994, IEEE. ISBN 0-7803-2135-9 .
  34. ^ I. Denkhaus: Traffic information systems: enforceability and acceptance in the Federal Republic of Germany. Springer-Verlag, 2013, p. 311.
  35. ↑ a b B. Reuse, R. Vollmar (Ed.): Informatikforschung in Deutschland. Springer Science & Business Media, 2008, p. 159.