Supercharge Your Innovation With Domain-Expert AI Agents!

Robot mapping method based on continuous confidence distribution

A confidence level and robot technology, applied in the field of robot perception, can solve the problems of unbalanced accuracy and efficiency, without considering the dependence of map grid units, etc., and achieve the effect of improving the accuracy of mutual information

Active Publication Date: 2021-08-27
ZHEJIANG UNIV
View PDF8 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to solve the problem that the traditional mutual information method based on the occupancy grid map does not consider the dependence between the map grid units, and the mutual information based on the inversion sensing model will lead to the imbalance of accuracy and efficiency. Occupying the continuous confidence distribution of the grid map, an information-theoretic metric called confidence-augmented mutual information is proposed; based on a causal sensing model, the dependence of measurements between grid cells within the same measurement cone at each time step Perform explicit modeling to obtain the confidence-enhanced mutual information of each grid cell intersecting the map on each beam, approximate and update the confidence-enhanced mutual information of the entire region under the new measurement, and build a real-valued map

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Robot mapping method based on continuous confidence distribution
  • Robot mapping method based on continuous confidence distribution
  • Robot mapping method based on continuous confidence distribution

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0098] Example 1: Small-scale narrow, messy simulation environment

[0099] In this embodiment, the simulation robot is equipped with a two-dimensional omnidirectional laser radar, which can evenly emit 60 beams, and the maximum measurement distance z MAX =8m, a total of 1144 frames of scan data were collected on the preset trajectory. Such as figure 2 As shown, the size of the simulation environment is 33m×33m, and the map resolution is λ M = 0.33m. The results of the confidence-enhanced mutual information CRMI in this embodiment are as follows: image 3 As shown, the values ​​of MI are all in the range of [0,1]. It can be seen that the CRMI distinction of boundary, free area, fuzzy area and undetected area is very obvious, and the order is increasing. The CRMI around the boundary and the robot itself is the smallest (dark black), which is mainly due to the cumulative calculation of the vacancy value (ie, 1 minus the occupancy value) during the SCM calculation and the b...

Embodiment 2

[0103] Example 2: Office environment experiment

[0104] This example uses the public dataset Intel Research Lab (Seattle), which contains 910 lidar scans, where the map size is set to 30m×30m, and the map resolution is λ M =0.2m, the maximum measurement distance is set to z MAX =8m,λ z = 0.1 m. Figure 6 For the real map of the dataset, Figure 7 and Figure 8 Schematic diagrams of CRMI and OGMI in this region, respectively. It can be seen that CRMI describes more details of the cluttered environment, such as corners, small objects and rooms, etc., while OGMI is relatively vague in these areas, and shows "overconfidence" in corridors and walls, which will also lead to Robots using information controllers have strong repulsive behavior in these areas, which may cause the robot to quickly pass over the area, making it difficult to complete the task of map construction and exploration in this small, restricted area.

Embodiment 3

[0105] Example 3: Large-scale campus scene environment experiment

[0106] In this embodiment, the public dataset Freiburg Campus containing 2008 laser radar scans is used. The scene is a large-scale unstructured outdoor environment, such as Figure 9 shown on the satellite map. In this embodiment, the map size is set to 250m×250m, and the map resolution is λ M =0.33m, the maximum measurement distance is set to z MAX = 50m. Such as Figure 10 As shown, CRMI exhibits similar performance to Embodiments 1 and 2, especially in the estimation of blurred areas, cluttered areas and undetected areas, which is more conservative, and at the same time, it shows refined estimation for small objects. The time consumption comparison in Table 2 also proves that the present invention can improve the estimation accuracy of MI without significantly increasing the computational complexity, which shows the superiority of the present invention.

[0107] Table 2 Time consumption comparison (MI...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a robot mapping method based on continuous confidence distribution, which comprises the following steps of: initializing an occupied grid map and confidence distribution, and initializing a beam hybrid sensing model; calculating the continuous confidence distribution of the map based on the distance measurement information of the airborne sensor; aiming at the problem that the traditional mutual information based on occupied grid maps does not consider the dependency between map grid units, so that the mutual information estimation precision and the operation efficiency are unbalanced, the invention provides mutual information measurement based on continuous confidence distribution, the measurement dependency between the grid units in the same measurement cone at the same moment is explicitly modeled, and the measurement accuracy is improved. The mutual information of each observed grid is obtained; Through an approximate calculation method, mutual information of the whole region under new measurement is updated, and a mutual information real value map is obtained. According to the mutual information mapping method, online implementation is allowed, and the robot can show a good exploration behavior for an unknown region and a fuzzy region in an unstructured and disordered environment.

Description

technical field [0001] The invention belongs to the technical field of robot perception, and mainly relates to a robot mapping method based on continuous confidence distribution. Background technique [0002] The purpose of the robot's active detection is to control the robot to draw more maps of unknown areas, while ensuring the accuracy of the map, time cost, travel distance, energy saving and other indicators. It has broad application prospects in military and civilian fields, such as ground detection of unknown planets and military reconnaissance. Currently, in the related art, the information-based detection method is more concerned, because this method allows faster detection, and it is easier to expand 3D scenes containing more irregularities and small boundaries. These methods utilize information-theoretic measures such as mutual information (MI) to construct reward functions and select optimal control strategies to minimize uncertainty in 2D and 3D maps. [0003] ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T17/05G06T7/73
CPCG06T17/05G06T7/73G06T2207/30244
Inventor 郑荣濠徐阳刘妹琴张森林
Owner ZHEJIANG UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com