UAV-Ground Vehicle Map Sharing for Distortion-Free Lane Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for autonomous vehicles face limitations in accurate lane detection and mapping due to field of view and view angle constraints of sensors, leading to distortions and inaccuracies when transforming images from a forward perspective to a birds-eye view, which affects reliability and processing efficiency.
Innovation Solution
The system shares sensor information between an unmanned aerial vehicle (UAV) and a ground vehicle, where the UAV captures birds-eye view data directly, eliminating the need for perspective transformation, and the ground vehicle uses this data for accurate lane detection and map generation, enhancing processing speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If perspective transformation is used to convert forward view sensor data to birds-eye view, then the field of view is expanded, but processing time increases and distortions are introduced
Solution Approach 1:
Instead of transforming forward view images to birds-eye view through complex perspective transformations, the system inverts the approach by directly capturing birds-eye view images using an aerial vehicle. This eliminates the need for computationally intensive perspective transformation while expanding the field of view, thereby resolving the contradiction between expanded coverage and processing time.
2Ease of operation
If perspective transformation is used to change the view angle of sensor data, then the viewing perspective is improved, but processing complexity and time increase
Solution Approach 1:
The system avoids complex perspective transformation processing by inverting the data collection approach. Instead of processing forward view images to achieve birds-eye view perspective, it directly captures birds-eye view images from an aerial vehicle, thereby improving viewing perspective while minimizing processing complexity.
3Ease of manufacture
If sensors are mounted on the ground vehicle, then the data collection is straightforward, but the field of view and view angle are limited
Solution Approach 1:
The system introduces an aerial vehicle as an intermediary to overcome the field of view limitations of ground-mounted sensors. The aerial vehicle captures birds-eye view images that provide expanded coverage, which are then shared with the ground vehicle for navigation, thereby resolving the contradiction between ease of data collection and expanded field of view.
4Ease of operation
If transformed data is used for navigation, then the perspective is improved, but reliability decreases due to distortions
Solution Approach 1:
The system eliminates reliability issues associated with transformed data by inverting the approach. Instead of transforming forward view images to achieve birds-eye view perspective, it directly captures undistorted birds-eye view images from an aerial vehicle. This provides both the improved perspective and high reliability needed for accurate navigation and lane detection.
Data Source
AI summary
Techniques are disclosed for sharing sensor information between multiple vehicles. A system for sharing sensor information between multiple vehicles, can include an aerial vehicle including a first computing device and first scanning sensor, and a ground vehicle including a second computing device and second scanning. The aerial vehicle can use the first scanning sensor to obtain first scanning data and transmit the first scanning data to the second computing device. The ground vehicle can receive the first scanning data from the first computing device, obtain second scanning data from the second scanning sensor, identify an overlapping portion of the first scanning data and the second scanning data based on at least one reference object in the scanning data, and execute a navigation control command based on one or more roadway objects identified in the overlapping portion of the first scanning data and the second scanning data.