amcl algorithm problem

tags: CSDN  ROS  Navigation

amcl understanding

Foreword

  First of all, this is my first blog post. I hope to help every ROS learner with my understanding. At the same time, as a beginner, I hope that all seniors can stop and exchange experiences. If there is a misunderstanding, I hope to point out.

text

    AMCL(adaptive Monte Carlo Localization)Adaptive Monte Carlo positioningFrom the sublimation and improvement of the MCL algorithm. So why go from MCL to AMCL?
So let's talk about the MCL algorithm. Monte Carlo positioning is applicable to both local positioning and global positioning problems. Although it is relatively young, it has become the mainstream algorithm in the positioning field. MCLIn the current formLocalization problem, but cannot recover from robot abduction or global positioning failure.When the position is acquired, incorrect particles elsewhere will gradually disappear.In a way, particles can only“survive”Near a single pose, if this pose happens to be incorrect, the algorithm cannot recover..
This problem is very important. In fact, any random algorithm, such as Monte Carlo algorithm, may accidentally discard all particles near the correct pose during the resampling step. When the number of particles is very small (M=50). Moreover, when the particles diffuse into the entire larger volume (such as the global positioning process), the importance of this problem can be fully displayed. Of course, this problem has been solved, AMCL's proposal is a solution to this problem. The AMCL algorithm randomly injects particles of injection when the robot is kidnapped. Compared with the MCL algorithm, the AMCL algorithm is exactly an "A"" , so what is this "A"? ""A"" is an acronym for adaptive, but personally think "A" is better understood as Augmented.
After clarifying the difference between MCL and AMCL, let's take a look at how AMCL is implemented. Please see the screenshot of the PPT made by the individual below.(Please look carefully at the content of ppt, there is relative knowledge to analyze, the text does not introduce too much)

(1) The difference between MCL algorithm and AMCL algorithm:

AMCL algorithm increases the short-term and long-term exponential filter attenuation rateαslow，αfast, in other words in MCLαslow，αfast is 0, and AMCL is not 0.



(2) The meaning of the four parameters



(3) Xt represents the set of M particles. Line 5 uses the motion model to obtain new poses from old particle sampling. Line 6 its importance weight is set according to the measurement model.



(4) The most important place in AMCL is the random sampling probability. (Look at the picture)



(5) motion_model uses sample_motion_model_velocity in Chapter 5 of the book "Probability Robot"



(6) The role of each parameter,



（7）measurement_model uses Chapter 6 of the book Probabilistic Robot landmark_model_known_correspondece



(8) The role of each part of the algorithm, and the input and output of the algorithm

Positioning display

    Borrowing an example from the book "Probablistic Robotics", a process of global positioning of the AMCL algorithm:




White dots represent: the actual position of the robot, and the red circle represents the average value from yourself (please carefully compare the chart, analyze the process, and gain a lot)

During the first mark detection, almost all particles are extracted according to this detection, as shown in the figureb shown. This step corresponds to the case where the short-term average of the measured probability is smaller than the long-term average of the measured probability.After multiple inspections, the particles are tightly wrapped around the early realityAround the robot, liked, and the short-term and long-term likelihood averages will increase. In this positioning stage, the robot is just trackingIts location, the observation likelihood is quite high, and only occasionally a small number of random particles are added.

When the robot is placed in another position, the measurement probability decreases. At this new location, the first mark detection has not triggered any additional particles because the smooth estimateWfast is still very high (e), After several marker tests at the new location,Wfast drops faster than Wslow, and more random particles are added (f, g). Finally, the robot positioning is successful.




Copyright notice: This article is an original article by bloggers and may not be reproduced without the permission of the blogger.

Intelligent Recommendation

ROS Navigation-----amcl node.cpp

Main function The main function of the main function is: Define a signal variable, manage the node Define the amclNode object 2. Construction of the amclNode object 1. There is a configuration related...

amcl source Interpretation

The first is the main function, the first step in the initialization operation of the conventional node amcl ros and instantiated. The second step uses the signal function: first parameter signal func...

Particle Filter AMCL

Outline Particle filter, i.e. Particle Filter, abbreviated PF. is placed in all the positioning program event data received laser sensor, a laser radar whether each received data, will determine what ...

amcl first look

Reprinted: About amcl The full name of amcl in English is adaptive Monte Carlo localization, which is actually an upgraded version of the Monte Carlo positioning method. The adaptive KLD method is use...

AMCL parameter description

translation AMCL introduction and parameter description 2017-03-17 17:09:52 Forrest-Z Reading 7040More Category column: ROS Package summary amcl is a probabilistic positioning system for robots to mov...

More Recommendation

amcl motion model

Odometer sport model Differential robot motion is decomposed into rotation, translation, and rotation. Define the Cartesian coordinate system, the right is the positive direction of the x axis, and th...

Run amcl on Pioneer3at

AMCL is a probabilistic positioning system for robots during two-dimensional movement. It uses adaptive Monte Carlo localization (or KLD sampling) and uses a particle filter to track the pose of the r...

Odom data processing in AMCL

Recently, I was watching the program of AMCL. I want to know how the configuration parameters of odom in the launch file operate. After reading it, make some notes.   1. AMCL's launch file about ...

ROS AMCL positioning

AMCL positioning is one of the more commonly used positioning methods in ROS. The specific principles are detailed in Chapter 8 of "Probability Robots". Here we only briefly record the probl...

Robot positioning (amcl) and navigation

Robot positioning (amcl) and navigation Description Simulation of positioning and navigation based on the successful construction of the map by gmapping amcl is a probabilistic positioning system in a...

Copyright  DMCA © 2018-2026 - All Rights Reserved - www.programmersought.com  User Notice

Top