Vehicle Position Estimation Using Contextual Sensor Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional approaches to determining a vehicle's position within its environment, such as SLAM techniques, require detailed and frequently updated maps, which are impractical for widespread implementation due to the need for extensive data collection and frequent updates, especially in dynamic environments like construction zones, and lack redundancy for accurate positioning.
Innovation Solution
The system generates a position estimate for a vehicle within a simplified map using data from various sensors like cameras, radars, and lidars, identifying objects and features to make positional inferences, and combines this information to create a probabilistic map indicating the likelihood of the vehicle's position, reducing the need for detailed maps and enabling accurate localization without relying on conventional SLAM techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional SLAM techniques are used to determine vehicle position, then positioning accuracy can be achieved, but the system requires detailed maps that need extensive data collection and frequent updates
Solution Approach 1:
The patent extracts only the necessary positional information from the environment (objects, features, lane markings) rather than requiring complete detailed maps. The system identifies and localizes specific salient objects and uses their known positions to infer vehicle location, eliminating the need for comprehensive map data while maintaining positioning accuracy.
Solution Approach 2:
Instead of requiring complete map coverage, the system uses partial information about the environment (identified objects and features) to achieve sufficient positioning accuracy. The approach uses only the necessary subset of environmental data needed for localization, reducing map data requirements while maintaining the required measurement precision.
2Measurement precision
If detailed maps are used for position estimation, then positioning can be accurate, but the maps require frequent updates in dynamic environments like construction zones
Solution Approach 1:
The system performs self-updating by automatically identifying and localizing current environmental objects and features during operation. Rather than relying on pre-existing maps that require manual updates, the system autonomously detects current environmental conditions (including changes like construction zones) and uses this real-time information to maintain positioning accuracy without external map updates.
Solution Approach 2:
The system prepares for environmental changes by continuously monitoring and identifying objects and features as they appear. This preliminary identification of current environmental elements allows the system to adapt to dynamic changes (such as construction zones) before they affect positioning, eliminating the need for frequent map updates.
3Measurement precision
If conventional SLAM techniques are used, then position estimation can be performed, but the system lacks redundancy for accurate positioning
Solution Approach 1:
The patent merges multiple data sources (sensor data identifying objects, features, lane markings, and traffic signals with pre-stored object information) to create a redundant positioning system. By combining data from multiple environmental references and using probabilistic maps that represent multiple possible positions, the system achieves both accuracy and redundancy for reliable positioning.
Solution Approach 2:
The system changes the representation of position from a single precise coordinate (as in conventional SLAM) to a probabilistic distribution representing multiple possible positions. This parameter change from deterministic to probabilistic positioning provides redundancy while maintaining accuracy, as the system can identify the most likely position from multiple possibilities.
Data Source
AI summary
Systems, methods, and non-transitory computer-readable media can receive data captured by one or more sensors associated with a vehicle. One or more objects in an environment of the vehicle can be identified based on the data captured by the one or more sensors. A position estimate of the vehicle can be generated within a known map based on one or more positional inferences pertaining to the vehicle, the one or more positional inferences pertaining to the vehicle being determined based on the one or more objects or features identified in the environment of the vehicle.