Autonomous Vehicle 2D Map Generation via Depth Laser Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicles face challenges in efficiently processing and utilizing the vast amounts of information from sensor systems to create accurate and useful maps for navigation, which hinders improved vehicle efficiencies and capabilities.
Innovation Solution
A method and system that involves obtaining images from cameras on the vehicle, creating a depth map using vehicle odometry information, generating a laser scan from the depth map, and constructing a two-dimensional map, utilizing various sensor systems such as accelerometers, wheel speed sensors, and global positioning systems to determine the camera pose, and using a controller to save and communicate the map for vehicle control systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If large amounts of information from wide array of sensor systems are processed to create accurate maps, then measurement precision and reliability are improved, but processing time and device complexity increase
Solution Approach 1:
The mapping process is segmented into distinct stages: obtaining images from cameras, creating depth maps from images and odometry, generating laser scans from depth maps, and constructing 2D maps from laser scans. This segmentation allows parallel processing and optimization at each stage, reducing overall processing time while maintaining map accuracy.
Solution Approach 2:
Vehicle odometry information is obtained and integrated into the depth map creation process before final map construction. This preliminary action of pre-processing odometry data and incorporating it early in the pipeline reduces the computational burden during real-time map generation, thereby reducing processing time without sacrificing precision.
2Measurement precision
If information from multiple sensor systems is integrated to determine camera pose and create depth maps, then measurement precision is improved, but device complexity increases
Solution Approach 1:
Data from multiple sensor systems (accelerometer, wheel speed sensor, wheel angle sensor, inertial measurement unit, global positioning system) are merged and integrated to determine camera pose. This combining of multiple sensor inputs creates a more robust and accurate pose estimation system while sharing the computational load across different sensor types.
Solution Approach 2:
The controller is designed to perform multiple functions: obtaining images from cameras, determining camera pose from various sensor systems, creating depth maps, generating laser scans, and constructing 2D maps. This multi-functional approach consolidates what could be separate complex systems into a single integrated controller, managing device complexity while maintaining measurement precision.
3Productivity
If real-time map creation is implemented for autonomous navigation, then productivity is improved, but processing burden and device complexity increase
Solution Approach 1:
The real-time mapping process is divided into sequential stages (image acquisition, depth map creation, laser scan generation, 2D map construction), allowing the controller to process one stage at a time. This segmentation enables real-time performance by breaking down the complex task into manageable computational steps that can be executed sequentially or in parallel.
Solution Approach 2:
The patent extracts and utilizes vehicle odometry information as a separate input source that is integrated into the depth map creation process. By extracting this pre-computed motion information from the vehicle's navigation system and incorporating it into the mapping pipeline, the controller reduces its own computational burden while maintaining real-time mapping capability.
Data Source
AI summary
A method and system for creating a map of an environment surrounding a vehicle includes a camera for obtaining images including objects within an environment from at least one camera mounted on the vehicle and a controller configured to create a depth map of the environment based on the images and vehicle odometry information. A laser scan of the depth map is created and used to create a two-dimensional map utilized for operating the vehicle.