Automated Vehicle Environmental Modeling With Factor-Graph SLAM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing environmental modeling methods for automated vehicles face challenges in maintaining accurate localization and maneuver planning due to deviations in high-definition maps caused by factors like vegetation or weather, and the formation of inconsistent navigation maps can lead to issues with tracking other road users and infrastructure dynamics.
Innovation Solution
Implementing locally consistent, factor-graph-based environmental modeling using the GTSAM algorithm (e.g., iSAM2) for sensor data fusion, which allows for the creation of a precise environmental model without jumps or errors within approximately 200-300 meters, enabling graph-based simultaneous localization and mapping (SLAM) and sensor data integration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high-definition map is used for environmental modeling, then localization accuracy is improved, but the system becomes vulnerable to deviations caused by vegetation or weather changes
Solution Approach 1:
The patent transitions from a static HD map approach to a dynamic environment model that is continuously updated at runtime using sensor data and factor graph optimization. This allows the system to adapt to changing environmental conditions (vegetation, weather) while maintaining localization accuracy, resolving the contradiction between initial precision and ongoing reliability.
Solution Approach 2:
The system performs preliminary actions by pre-processing sensor data into a factor graph structure that can be efficiently optimized. The factor graph is prepared in advance with all necessary sensor measurements and constraints, enabling rapid computation of the environment model when needed, thus maintaining both accuracy and reliability.
2Adaptability or versatility
If only a navigation map is present and environment modeling is done at runtime, then adaptability is improved, but local consistency may be compromised with jumps and errors
Solution Approach 1:
The factor graph optimization process continuously incorporates feedback from multiple sensors (cameras, LIDAR, IMU, odometry) to refine the environment model. This feedback mechanism ensures that the runtime-generated model remains locally consistent by constantly comparing new sensor data with the existing model and adjusting accordingly, preventing jumps and errors while maintaining adaptability.
Solution Approach 2:
The patent combines multiple data sources (infrastructure data, object data, trip data) into a composite environment model using factor graph optimization. This composite approach integrates heterogeneous sensor information with different levels of reliability, creating a robust and consistent environment model that maintains stability while adapting to runtime conditions.
3Measurement precision
If sensor data is continuously integrated into the factor graph, then model precision is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational process into distinct modules: data reading, factor graph construction, optimization, and model extraction. This segmentation allows each module to be optimized independently and enables parallel processing where possible, reducing overall computational complexity while maintaining model precision through systematic data integration.
Data Source
AI summary
A method for operating a vehicle for highly automated driving. The method includes a step of reading sensor data that comprise trip data of the vehicle from at least one acceleration sensor, at least one position sensor, and a velocity sensor, infrastructure data of infrastructure elements in a predefined environment of the vehicle from at least one environmental sensor, and object data of recognized traffic objects in the predefined environment from the environmental sensor. An environmental model for behavior planning and maneuver planning of the vehicle within the predefined environment is determined. The environmental model is determined by simultaneous localization and mapping using the sensor data and a factor graph into which the sensor data are integrated as factors. The environmental model is output to an interface to a planning device for behavior planning and maneuver planning of the vehicle.