Indoor Robot Mapping with Loop Closure Pose Graph Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic mapping technologies suffer from error accumulation in dynamic environments due to the lack of loop closure optimization, leading to inaccuracies in indoor map creation, particularly in sites like disaster areas where situations change frequently.
Innovation Solution
Implementing loop closure techniques using a closed-loop trajectory with pose graph optimization and condition-based loop route selection, employing sensors like LiDAR, IMU, and vision sensors to refine indoor maps by updating and optimizing odometry and trajectory data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a one-way open loop method is used for odometry estimation and trajectory optimization, then the system can operate with simpler processing, but errors accumulate over time leading to reduced mapping precision
Solution Approach 1:
The patent implements loop closure detection and optimization by detecting when the mobile robot returns to a previously visited location and using this feedback information to correct accumulated odometry errors. The system compares current sensor data with historical data from the same location and adjusts the trajectory accordingly, transforming the open-loop system into a closed-loop system that continuously corrects its own errors.
Solution Approach 2:
The patent performs preliminary odometry estimation and local map generation before loop closure detection. By pre-processing the trajectory data and creating initial maps, the system prepares the necessary foundation for subsequent error correction, allowing the complex optimization to be applied selectively rather than continuously.
2Measurement precision
If loop closure optimization is implemented to reduce error accumulation, then mapping precision is improved, but the system complexity and computational load increase
Solution Approach 1:
The patent applies loop closure optimization selectively rather than continuously. The system detects loop closure events and performs optimization only when such events are detected, rather than applying complex optimization algorithms at all times. This partial application of the optimization reduces the overall computational burden while still achieving error correction when needed.
Solution Approach 2:
The patent divides the mapping process into distinct segments: initial map generation using open-loop odometry, loop closure detection, and localized optimization. By segmenting the overall process, the system can use simple methods for most of the time and only invoke complex optimization when loop closure is detected, reducing average system complexity.
3Reliability
If the robot continuously updates maps using open-loop trajectories, then the mapping process is simpler to implement, but errors accumulate in dynamic environments where situations change
Solution Approach 1:
The patent uses loop closure detection as a feedback mechanism to identify when the robot has returned to a previously visited location in a dynamic environment. This feedback triggers a recalculation and update of the local map using the corrected closed-loop trajectory, ensuring that the map reflects the current state of the environment rather than accumulating errors from outdated open-loop trajectories.
Data Source
AI summary
The present disclosure relates to an operation method of a computing device for performing a method for producing precise indoor maps using loop closing, including the steps of moving a robot to a first location, wherein the robot collects images and sensor data for precise map production by using a sensor module including at least one of a LiDAR sensor, an IMU sensor; forming a first closed-loop trajectory with an optimized pose graph by odometry estimation from the first location to create a first map using the movement of the robot; updating at least part of a second map generation trajectory and sensor data corresponding to a current trajectory of the robot by using at least one of trajectory information and sensor data of the first closed-loop trajectory; and outputting an updated second map based on the updated second map generation trajectory and sensor data.