Image-Space Trajectory Planning for Depth-Uncertain Autonomous Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating through environments with uncertain depth estimates, particularly due to complex shapes like trees with intermittent foliage, which can lead to unreliable depth measurements and increased collision risks.
Innovation Solution
The implementation of image space motion planning techniques that identify regions with low confidence depth estimates and associate them with higher collision risks, optimizing the vehicle's trajectory to avoid these areas by projecting the 3D trajectory into the image space and adjusting the path to minimize risk, using cost functions and machine learning to assess the level of risk associated with different regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the autonomous vehicle uses visual odometry to estimate position and orientation based on captured images, then the navigation system can guide the vehicle through the physical environment, but the depth estimates become unreliable in regions with complex shapes like trees with intermittent foliage, leading to increased collision risks
Solution Approach 1:
The patent divides the image into multiple depth regions (first depth region with reliable depth estimates and second depth region with unreliable depth estimates) based on depth uncertainty. This segmentation allows the system to treat different regions differently, avoiding high-uncertainty areas while navigating through low-uncertainty areas, thereby resolving the contradiction between using visual odometry for navigation and avoiding collisions in regions with complex shapes.
2Object-affected harmful factors
If the autonomous vehicle optimizes its trajectory to avoid regions with low confidence depth estimates, then the collision risk is reduced, but the navigation path may become longer or more complex
Solution Approach 1:
The patent dynamically adjusts the vehicle's trajectory by continuously updating the planned path based on real-time depth uncertainty maps. The motion planning system recalculates the optimal path considering the current positions of both low-uncertainty and high-uncertainty regions, allowing the vehicle to adaptively navigate around unreliable depth areas while maintaining efficient progress toward the destination.
3Measurement precision
If the autonomous vehicle uses multiple image capture devices to improve depth estimation accuracy, then the reliability of depth measurements increases, but the device complexity and computational load increase
Solution Approach 1:
The patent introduces depth uncertainty maps as an intermediary representation that simplifies the complex multi-device image data into a usable format for motion planning. Instead of directly processing raw images from multiple capture devices, the system first generates depth maps and then creates uncertainty maps that highlight reliable versus unreliable regions, serving as a mediator between the complex sensor data and the navigation decision-making process.
Data Source
AI summary
An autonomous vehicle that is equipped with image capture devices can use information gathered from the image capture devices to plan a future three-dimensional (3D) trajectory through a physical environment. To this end, a technique is described for image-space based motion planning. In an embodiment, a planned 3D trajectory is projected into an image-space of an image captured by the autonomous vehicle. The planned 3D trajectory is then optimized according to a cost function derived from information (e.g., depth estimates) in the captured image. The cost function associates higher cost values with identified regions of the captured image that are associated with areas of the physical environment into which travel is risky or otherwise undesirable. The autonomous vehicle is thereby encouraged to avoid these areas while satisfying other motion planning objectives.