Autonomous Machine Vision Sensor Selection for Low-Load Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous ground maintenance machines, such as lawn mowers, face inefficiencies in navigation due to high utilization of vision-related computing resources and limited flexibility in using vision systems for various tasks, particularly in defining work regions and correcting machine positions within these regions.
Innovation Solution
The implementation of a method that utilizes multiple vision sensors with distinct fields of view, where one sensor is designated as a localization vision sensor based on feature matches with a pre-identified 3D point cloud, and visual odometry data is collected to correct the machine's pose, reducing computing resource usage and enhancing flexibility in camera utilization for tasks like localization, object classification, and visual odometry.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple vision sensors are used for navigation and localization, then measurement precision and reliability are improved, but computing resource utilization increases
Solution Approach 1:
The patent divides the vision sensing function into two distinct segments: a localization vision sensor dedicated to pose determination and other vision sensors used for additional tasks. This segmentation allows the localization sensor to focus computational resources exclusively on accurate pose determination through 3DPC feature matching, while other sensors handle secondary functions, thereby maintaining measurement precision without requiring all sensors to operate at full computational capacity simultaneously.
Solution Approach 2:
The patent implements multi-functionality by enabling non-localization vision sensors to serve multiple purposes: primary navigation tasks, object classification, and visual odometry. This universality reduces the need for dedicated single-function sensors, allowing the system to achieve reliable navigation with fewer sensors operating at high computational intensity, thus reducing overall computing resource utilization while maintaining precision.
2Adaptability or versatility
If vision sensors are used for multiple tasks simultaneously, then adaptability is improved, but device complexity increases
Solution Approach 1:
The patent implements dynamic sensor role assignment where the localization vision sensor designation can change based on operational context. The system dynamically selects which sensor serves localization functions by evaluating real-time factors such as sensor availability, task requirements, and environmental conditions. This dynamic allocation simplifies device complexity by providing clear role differentiation at any given moment while maintaining adaptability across different operating scenarios.
Solution Approach 2:
The vision sensor system performs self-service through automated task allocation and sensor selection. The controller automatically determines which sensor should serve as the localization sensor based on predefined criteria and current operational state, without requiring manual configuration or complex external management. This self-service mechanism reduces device complexity by eliminating the need for intricate sensor management interfaces while preserving adaptability through automated decision-making.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Vision systems for autonomous machines and methods of using same during machine localization are provided. Exemplary systems and methods may reduce computing resources needed to perform vision-based localization by selecting the most appropriate camera from two or more cameras, and optionally selecting only a portion of the selected camera's field of view, from which to perform vision-based location correction. Other embodiments may provide camera lens coverings that maintain optical clarity while operating within debris-filled environments.