File Exchange

image thumbnail


version (15.6 KB) by Ruslan Masinjila
Multirobot Localization Using Extended Kalman Filter


Updated 15 Feb 2017

GitHub view license on GitHub

Multirobot Localization Using Extendend Kalman Filter.
>Change number of robots, simulation length and number of runs
A group of N robots with known but uncertain initial poses move randomly in an open, obstacle-free environment. As some of robots (one or more) move, the rest of robots (at least one), remain stationary and act as landmarks to the moving robots, and vice versa.
More information is available on Chapter 3 (page 21 to 58)

For each robot, the following assumptions and models were made:

motion model: unicycle (2-wheeled robot).
measurement model: relative distance (rho) and relative angle (phi)
encoder noise: gaussian and linearly proportional to the distance moved by the wheels.
range sensor noise: gaussian in rho and phi

February 4th 2017 [version1]
> general case for multirobot localization involving N robots where N>=2

February 5th 2017 [version2]
> allows the simulation to be repeated numRuns times under the same initial conditions, control inputs, and sequence of movements.
> calculates and displays estimated results averaged over numRuns times.
> computes and displays ANEES averaged over numRuns.

February 6th 2017 [version3]
> compute and display absolute errors in X and Y coordinates of all robots.

> display actual and estimated 2D error ellipses around estimated robot poses.
> add covariance inflation index for tuning the ekf algorithm.

Cite As

Ruslan Masinjila (2020). multirobot_ekf_localization (, GitHub. Retrieved .

Comments and Ratings (0)


displays graphs for ground truths, ekf estimates, encoder estimates, anees, and absolute errors in x and y coordinates

New Version (v2.0)

MATLAB Release Compatibility
Created with R2014a
Compatible with any release
Platform Compatibility
Windows macOS Linux