VEHICLE WITH VISUAL SIMULTANEOUS LOCALIZATION AND MAPPING SYSTEM AND DYNAMIC OUTLIER REJECTION FUNCTIONALITY

Smart overview of the Invention

A Visual Simultaneous Localization and Mapping (V-SLAM) system is integrated into a mobile host, such as a vehicle, to enhance navigation accuracy. This system utilizes a sensor suite and a controller to gather raw input data, estimate the host's parameters, and calculate 3D coordinates of environmental features. A key function of the controller is to identify and filter out outliers from these feature map points using a novel approach involving dynamic covariance scaling and condition numbers, thereby refining the data used by the navigation system.

System Composition

The V-SLAM system comprises a sensor suite and a controller. The sensor suite collects data from various sources, including cameras for visual input and potentially a GPS receiver for position data. The controller processes this data through a processor and memory containing specific instructions. It estimates the vehicle's position and orientation, identifies outliers in the data using linear algebra properties, and dynamically filters these outliers to produce a reliable set of feature map points for navigation.

Outlier Rejection Mechanism

The system's effectiveness hinges on its ability to reject ill-conditioned feature map points that could compromise navigation accuracy. By utilizing condition numbers as a thresholding mechanism, the system identifies points that contribute to an unstable matrix. These points are dynamically scaled or discarded based on their condition number, ensuring that only well-conditioned data is retained for further processing. This process significantly enhances the system's robustness, especially in environments where GPS signals are unreliable.

Functionality and Benefits

Execution of the system's instructions enables the controller to refine the navigation data by filtering out inconsistent feature map points. This leads to improved localization accuracy and operational stability. The system is particularly beneficial in challenging environments such as urban canyons where GPS signals may be obstructed. By dynamically adjusting the scaling of moderately ill-conditioned points, the system maintains a high level of accuracy and reliability in its navigational outputs.

Implementation and Application

The V-SLAM system is adaptable to various mobile hosts, including vehicles, and integrates seamlessly with their navigation systems. The sensor suite may include additional components like an inertial measurement unit (IMU) to further enhance pose estimation. The controller's ability to dynamically filter and transmit refined map points to the navigation system ensures precise control and navigation capabilities. This technology offers significant advancements in autonomous vehicle navigation, contributing to safer and more efficient travel.