Guaranteed Bounds for Vehicle Motion State Estimates for automated driving
Purpose and goal: Reliable estimates of vehicle condition (position and movement) are crucial for our success with electrification and automation of long vehicle combinations. This project studies how to improve the uncertainty description of the condition of vehicles so that we can realize safe automated driving in complex traffic environments.
The project will
1) identify how, for example, sensors can be modeled to improve the state estimates,
2) develop new estimation methods with a focus on uncertainty descriptions and
3) explore how machine learning can be applied in this context.
Expected results and effects: The project is expected to give the following results:
- Guidelines for how autonomous systems should be designed to get a good uncertainty description
- Models and methods to accurately describe the uncertainty in the system (model-based and trained from data)
- Uncertainty descriptions in the system capturing uncertainty making it manageable for subsequent decision algorithms
- Data for further research on motion state estimates of heavy vehicles.
- Licentiate (and doctoral dissertation)
- Demonstration (together with "MULTITRAILER" no. 2020-05144) - Competence build-up
Approach and implementation: The project has been divided into six work packages (WP´s):
WP1: Complete vehicle requirements, what the system should be able to do etc
WP2: System design and analysis - 2.1 System design - 2.2 System analysis
WP3: Safe vehicle state estimation - 3.1 System design - 3.2 Sensor models - 3.3 Vehicle models - 3.4 Secure estimates
WP4: Functional Design and decision-making
WP5: Testing, validation and data collection Testing, validation, and data collection will be done using full-scale vehicles equipped with various sensors etc.
Fredrik Von Corswant (contact)
Project ID: 2021-02570
Funding Chalmers participation during 2021–2024