Central system for supporting automated vehicle testing and operation
The project is focusing on realizing a transport system to develop and demonstrate a holistic solution to support and operate autonomous vehicles in cooperation with infrastructure elements. The project targets to implement the system for testing purposes in the beginning, but should evolve for transport operation and control. In today’s mainstream for operation of highly automated vehicles the main environmental model for vehicle guidance is generated based on the ego vehicle’s perception system and eventually extended with infrastructure’s and other vehicle’s information. Our proposed solution builds up the global environment model externally in a cloud and supports the individual vehicles with a comprehensive world model or even can control them. The system will collect all information from both vehicle and infrastructure side and fuse them together in a cloud based, real time digital twin.
TU Wien is researching on vehicle state and parameter estimation as well as on controllability of critical driving conditions with several actuators. TU Wien and STARD can build on a preceding FFG project on predictable handling behavior of vehicles due to torque vectoring control on several road surfaces, where the focus was put on simulation studies, and recently started a cooperation on parameter estimation with BME. Based on these previous works, TU Wien puts a strong research focus on automated driving at the limits of handling to fully exploit the safety potential of today’s over-actuated vehicles in Central Europe’s climate (and road) conditions.
To expand the effectiveness of the vehicle response to track guiding commands as well as to enhance the predictability of the vehicle behavior at given sensor/actuator configurations and failures, a combined controllability and observability measure will be developed. This measure will be available in the cloud to enhance the cloud-based control methodology by regarding the actual capabilities of the automatic vehicle considering the control constraints resulting from the dynamic cloud map. Further, from this measure, an actuator allocation control strategy will be developed that may be utilized by the cloud guidance controller, which will be developed under the lead of the Hungarian partner BME. In addition, from effect-based methods, a locally estimated friction potential will be provided to the cloud. The effectiveness of the proposed approach will be demonstrated utilizing the over-actuated demonstrator vehicle.
The challenges to the Austrian consortium of the overall project are to expand and enhance a controllability and observability measure to allow for most effective/efficient system actuation at several combinations of sensor/actuators sets to ensure redundant and safe automatic vehicle control, since the actual vehicle state as well as road friction potential may alter respective measures.
- Qualitative measure: A cloud-based control methodology will be available that allows to perform automated driving scenarios utilizing an over-actuated vehicle at considering actual vehicle state, traffic, weather conditions etc.
- Quantitative measure: The developed methodology will be demonstrated on a demonstrator vehicle utilizing cloud control at performing one path tracking maneuver and one accident avoidance maneuver on the proving ground ZalaZONE.
- STARD part of Stohl Group GmbH
- BME - Automated Drive
- TU Graz
- JOANNEUM RESEARCH Forschungsgesellschaft mbH
- TOM Robotics GmbH
- Virtual Vehicle Research GmbH
- BOSCH HU
- Telekom HU
- Central System: https://projekte.ffg.at/projekt/4105771
September 2021 - August 2024