Autonomous Vehicle World Model With Hypothetical Dark Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicle systems face challenges in making safe driving decisions due to the inability to perceive and account for objects that are not within their sensor range, leading to potential collisions with unknown or unobserved obstacles.
Innovation Solution
The system generates a world model that includes hypothetical 'dark objects' in unperceived areas, using data from sensors and road network databases to infer the presence and attributes of these objects, such as vehicles, pedestrians, and animals, and updates vehicle trajectories to account for potential risks from these unknown entities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the AV system relies only on sensor-perceived objects for driving decisions, then the system complexity remains low and processing is straightforward, but the safety deteriorates due to inability to detect unknown objects beyond sensor range
Solution Approach 1:
The system performs preliminary actions by generating hypothetical dark objects in unperceived areas before actual collisions can occur. The planning module proactively creates virtual objects in regions beyond sensor range based on probabilistic reasoning, allowing the vehicle to plan avoidance trajectories in advance rather than reacting to actual threats.
Solution Approach 2:
The patent introduces an intermediary layer between sensor perception and decision-making. The dark object generation module acts as a mediator that bridges the gap between known perceived objects and unknown unperceived objects, creating hypothetical representations that enable safer planning without requiring direct sensor detection of all objects.
2Loss of information
If the AV system expands sensor coverage to detect more objects, then the ability to perceive environment improves, but the energy consumption and computational load increase
Solution Approach 1:
Instead of expanding sensor coverage to physically detect all objects, the system performs preliminary computational action by generating hypothetical dark objects in unperceived regions. This approach recovers information about potential objects through reasoning rather than through additional sensor energy consumption.
Solution Approach 2:
The system creates virtual copies of potential objects (dark objects) in unperceived areas based on probabilistic reasoning about what objects might exist there. These copied representations allow the planning system to account for potential hazards without requiring physical sensor detection, thus avoiding additional energy expenditure while still improving environment perception.
3Reliability
If the AV system generates multiple hypothetical dark objects in unperceived areas, then the coverage of potential hazards improves, but the computational complexity and processing time increase
Solution Approach 1:
The system generates multiple hypothetical dark objects in advance during the planning phase, before actual navigation decisions are made. This preliminary generation of potential hazard scenarios allows for comprehensive safety consideration without adding time-critical computational burden during real-time control.
Solution Approach 2:
The system generates an excessive number of hypothetical dark objects beyond what would be minimally necessary, covering all unperceived areas with potential objects. This partial/excessive coverage ensures that no potential hazard is missed, and the planning system can safely navigate by considering all possible scenarios, with the understanding that not all generated objects will be actual threats.
Data Source
AI summary
Among other things, a model is maintained of an environment of a vehicle. A hypothetical object in the environment that cannot be perceived by sensors of the vehicle is included in the model.