I develop this package every weekend.

Work in progress

90%

Install Requirements

$ git clone https://github.com/yukkysaito/amcl_3d.git <catkin_ws/src path>
$ rosdep install --from-paths <amcl_3d path> -y

TODO

• add prediction model (particle filter)

Done

• design interface (particle filter)
• make ros package template
• add basic resample function (particle filter)
• add init function(particle filter)
• build check (travis ci)
• resampling timing based on ESS (effective sample size)
• add simple likelihood model for lidar measurement (particle filter)
• implement argumented mcl
• implement kld sampling

Reference

English Reference

General

• http://www.probabilistic-robotics.org/

Effective Sampling Size

• http://www.cns.nyu.edu/~eorhan/notes/particle-filtering.pdf

Avoidance Numerical Underflow

• http://www.maths.lu.se/fileadmin/maths/forskning_research/InferPartObsProcess/particlemethods.pdf

Japanese Reference

• http://lab.cntl.kyutech.ac.jp/~nishida/lecture/psc/no11.pdf
• https://gihyo.jp/book/2018/978-4-7741-9646-6
• https://qiita.com/MoriKen/items/dfb6eb168649873589f0