Autonomous Vehicle Occlusion Mapping for Hidden Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in detecting objects in occluded areas due to sensor obstruction by parts of the vehicle, which can lead to navigation difficulties and increased collision risks.
Innovation Solution
Systems and methods that process sensor data using machine learning models and configuration data to identify occluded areas and determine the presence of objects within these areas, allowing for adjustments in vehicle trajectory or sensor positioning to enhance visibility and avoid collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensors are positioned on the vehicle to detect objects, then object detection capability is improved, but sensor obstruction by vehicle parts occurs reducing detection accuracy
Solution Approach 1:
The patent projects 2D sensor data into 3D space to create a 3D point cloud representation of the environment. This dimensional transformation allows the system to reconstruct occluded areas by filling gaps in the 3D space where points are missing, effectively compensating for sensor obstruction by vehicle parts
Solution Approach 2:
The system creates a digital copy of the environment through point cloud generation and occlusion modeling. By generating a virtual representation of occluded areas based on detected objects' trajectories and environmental data, the system compensates for physical sensor limitations without adding physical sensors
2Reliability
If the vehicle structure is maintained for operational requirements, then vehicle functionality is preserved, but occluded areas increase reducing safety
Solution Approach 1:
The system performs preliminary identification of occluded areas by analyzing sensor data patterns and object trajectories before actual collisions or detection failures occur. By pre-mapping occlusion zones and predicting potential hazards in these areas, the system can take preventive navigation actions
Solution Approach 2:
The patent introduces an intermediary occlusion identification system that mediates between the physical vehicle structure and the navigation decision-making process. This intermediary layer processes sensor data to infer information about occluded areas, translating physical limitations into actionable navigation intelligence
3Measurement precision
If sensors are added to reduce occlusion, then detection coverage is improved, but device complexity increases
Solution Approach 1:
The patent replaces physical mechanical sensor solutions with computational methods. Instead of adding more physical sensors to cover occluded areas, the system uses algorithms to process existing sensor data, generate point clouds, model occlusions, and infer environmental information computationally
Data Source
AI summary
In various examples, identifying occluded areas of environments for autonomous systems and applications is described herein. For instance, systems and methods described herein may process sensor data generated using one or more sensors of a machine to determine one or more areas of an environment that are occluded by at least a part of the machine. In some examples, determining the area(s) of the environment may be based at least on processing the sensor data using one or more machine learning models, processing the sensor data with respect to configuration data, and/or using any other technique. The systems and methods may then determine whether one or more objects are located within the occluded area(s) of the environment. Additionally, the systems and methods may then cause the machine to perform one or more operations based at least on whether the object(s) is located within the occluded area(s).