3D Model Generation Using Camera and LIDAR Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating 3D models of environments, such as those around robots or vehicles, do not produce high-quality models efficiently, especially in areas where extrinsic positioning systems like GPS are unavailable.
Innovation Solution
A method using a combination of a first sensor, typically a camera, and at least one LIDAR sensor, where the sensors are controlled by independent clocks, to determine the trajectory and scan the environment, processing the data using statistical methods to match LIDAR returns with the trajectory, creating a representation of the environment without relying on extrinsic positioning systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If prior art methods are used to generate 3D models, then the process is simpler, but the model quality is insufficient
Solution Approach 1:
The patent combines data from multiple independent sensors (LIDAR and camera systems with separate clocks) into a unified 3D model. The synchronization module merges timing information from different clock sources to correlate LIDAR range data with camera image data, achieving high-quality models through multi-source data fusion rather than relying on a single sensor system.
Solution Approach 2:
The patent introduces a synchronization module as an intermediary component that processes timing information from independent sensor clocks. This mediator correlates the timing data from LIDAR and camera systems, enabling accurate temporal alignment of measurements without requiring the sensors to share a common clock, thus resolving the complexity while maintaining precision.
2Adaptability or versatility
If extrinsic positioning systems like GPS are used, then positioning is easier, but the system cannot operate in areas where GPS is unavailable
Solution Approach 1:
The patent extracts the positioning function from extrinsic GPS-dependent systems and implements it through intrinsic sensor-based relative positioning. By using LIDAR range measurements and camera imagery processed through structure-from-motion algorithms, the system determines trajectory and environmental geometry independently of external positioning satellites, enabling operation in GPS-denied environments.
Solution Approach 2:
The system performs self-positioning and self-mapping by using its own sensors (LIDAR and cameras) to simultaneously determine its trajectory through the environment and create 3D models of that environment. This self-contained approach eliminates dependency on external positioning infrastructure, allowing the monitoring unit to operate autonomously in any environment.
3Reliability
If independent clocks are used for each sensor, then sensor independence is maintained, but timing synchronization becomes more difficult
Solution Approach 1:
The patent replaces the mechanical approach of sharing a common clock among sensors with an information-based solution. Each sensor maintains its own independent clock, and the synchronization module uses software algorithms to process and correlate the timing information from these independent sources. This substitution of mechanical synchronization with computational correlation maintains sensor independence while achieving precise temporal alignment of measurements.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach generates higher quality 3D models of environments, enabling effective navigation and assessment, particularly in areas inaccessible to GPS, by combining camera and LIDAR data to create accurate point clouds or other 3D reconstructions.
Implementation Method 1
at least one LIDAR sensor, wherein the first and second sensors may be provided at a fixed orientation relative to one another
Implementation Method 2
scanning the environment with the LIDAR sensor, recording the returns from the LIDAR sensor
Data Source
AI summary
Generating a 3D reconstruction of an environment around a monitoring-unit as that monitoring-unit is moved through the environment: a) providing at least a camera and a LIDAR sensor, each being controlled by independent clocks; b) using the camera to determine the trajectory of the monitoring-unit and determining a first time series using the clock of the camera, where the first time series details when the monitoring-unit was at predetermined points of the trajectory; c) recording the returns from the LIDAR sensor and determining a second time series using the clock of the LIDAR sensor, where the second time series details when each scan from the LIDAR was taken; d) using a timer to relate the first and second series in order to match the return from the LIDAR sensor to the point on the trajectory at which the return was received; and e) creating the 3D reconstruction based upon the LIDAR returns using information from the two time series.