Cloud Trajectory Map Alignment for High-Precision SLAM Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory map generation schemes for robots and UAVs face challenges in producing high-precision maps due to limitations in monocular SLAM, binocular SLAM, and other vision-based methods, which result in scale drift and difficulty in achieving data synchronization, leading to rough and inaccurate maps.
Innovation Solution
A cloud-based trajectory map generation method that acquires data from multiple mapping schemes, aligns and interpolates the data to determine an optimal transformation relationship, and generates a high-precision map by solving a residual equation, using modules for data acquisition, initialization, interpolation, optimization, and map generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple SLAM schemes or repeated SLAM schemes are used to generate trajectory maps for the same region, then map precision can be improved, but data synchronization becomes difficult due to randomness of feature points and key frames and differences in timestamp systems
Solution Approach 1:
The patent introduces a cloud-based server as an intermediary to receive trajectory map data from multiple robots, perform centralized alignment and optimization calculations, and return optimized transformation relationships. This mediator handles the complex data synchronization task that would be difficult for individual robots to accomplish independently, resolving the contradiction between improved map precision and synchronization complexity.
Solution Approach 2:
The system performs preliminary alignment using map-initializing treatment to obtain initial transformation relationships before conducting detailed optimization. This preliminary action prepares the data in advance, making subsequent synchronization and optimization more efficient and manageable.
2Device complexity
If monocular SLAM scheme is used, then system complexity is reduced, but real-scale map generation is impossible and scale drift occurs
Solution Approach 1:
The patent combines trajectory map data from multiple robots that may have been generated using different SLAM schemes. By merging multiple data sources and performing joint optimization, the system achieves accurate scale information without requiring each individual robot to use complex schemes like binocular SLAM or VIO.
Solution Approach 2:
The system uses feedback from multiple trajectory maps to iteratively optimize transformation relationships. The cloud server receives data from multiple robots, performs alignment and optimization, and uses the feedback from multiple sources to correct scale drift and improve overall map accuracy.
3Measurement precision
If binocular SLAM scheme is used, then real-scale map generation is achieved, but high-precision mapping in outdoor large-depth scenes is difficult due to baseline length limitations
Solution Approach 1:
The patent transitions from relying on individual robot baseline measurements (3D spatial limitation) to utilizing the temporal and spatial dimensions of multiple robots' trajectories. By combining trajectory data from multiple robots moving through the same region, the system achieves accurate large-depth mapping without being constrained by individual baseline lengths.
4Measurement precision
If trajectory map data from different mapping schemes are combined, then map precision is improved, but data alignment and synchronization become complex
Solution Approach 1:
The cloud-based server acts as a mediator that receives trajectory map data from multiple robots using different mapping schemes, performs centralized alignment and optimization, and returns coordinated transformation relationships. This centralized intermediary handles the complexity of aligning data from different sources, schemes, and timestamp systems.
Solution Approach 2:
The system changes the parameters of transformation relationships through iterative optimization. By adjusting transformation parameters based on residual equations and optimization algorithms, the system aligns data from different mapping schemes while improving overall precision.
Data Source
AI summary
A cloud-based trajectory map generation method, device, apparatus and application. The method includes: acquiring first trajectory map data and second trajectory map data; performing a map-initializing treatment to align the two trajectory map data to obtain an initial value of a transformation relationship therebetween; interpolating the two trajectory map data to obtain corresponding data at preset interpolation points thereto; determining a residual equation based on the corresponding data at the preset interpolation points and the initial value of the transformation relationship to obtain an optimal solution of the transformation relationship based thereon; and generating a trajectory map based on the optimal solution of the transformation relationship. The trajectory map data are obtained using two different mapping schemes over a same moving trajectory, or using a same mapping scheme over two substantially identical moving trajectories, within a same region, which are optimized to thereby obtain a high-precision trajectory map.