Mapping and Localization

Lidar scan mapping, and particle filter localization

Create maps of environments using occupancy grids and localize using a sampling-based recursive Bayesian estimation algorithm using lidar sensor data from your robot.

Functions

expand all

Mapping

`binaryOccupancyMap`	Create occupancy grid with binary values
`getOccupancy`	Get occupancy value of locations
`inflate`	Inflate each occupied location
`move`	Move map in world frame (Since R2019b)
`occupancyMatrix`	Convert occupancy grid to matrix
`raycast`	Compute cell indices along a ray (Since R2019b)

Laser Scans

`lidarScan`	Create object for storing 2-D lidar scan (Since R2019b)
`plot`	Display laser or lidar scan readings
`removeInvalidData`	Remove invalid range and angle data
`transformScan`	Transform laser scan based on relative pose

Localization

`stateEstimatorPF`	Create particle filter state estimator
`initialize`	Initialize the state of the particle filter
`predict`	Predict state of robot in next time step
`correct`	Adjust state estimate based on sensor measurement
`getStateEstimate`	Extract best state estimate and covariance from particles

Topics

Mapping

Occupancy Grids
Details of occupancy grid functionality and map structure.

State Estimation

Particle Filter Parameters
To use the stateEstimatorPF particle filter, you must specify parameters such as the number of particles, the initial particle location, and the state estimation method.
Particle Filter Workflow
A particle filter is a recursive, Bayesian state estimator that uses discrete particles to approximate the posterior distribution of the estimated state.

Featured Examples

Mapping with Known Poses

Create a map of an environment using range sensor readings and robot poses for a differential drive robot. You create a map from range sensor readings that are simulated using the rangeSensor object. The differentialDriveKinematics motion model simulates driving the robot around the room based on velocity commands. The rangeSensor gives range readings based on the pose of the robot as it follows the path.

Open Live Script

Track a Car-Like Robot Using Particle Filter

Particle filter is a sampling-based recursive Bayesian estimation algorithm, which is implemented in the stateEstimatorPF object. In this example, a remote-controlled car-like robot is being tracked in the outdoor environment. The robot pose measurement is provided by an on-board GPS, which is noisy. There are known motion commands sent to the robot, but the robot will not execute the exact commanded motion due to mechanical slack or model inaccuracy. This example will show how to use stateEstimatorPF to reduce the effects of noise in the measurement data and get a more accurate estimation of the pose of the robot. The kinematic model of a car-like robot is described by the following non-linear system. The particle filter is ideally suited for estimating the state of such kind of systems, as it can deal with the inherent non-linearities.

Open Live Script

Avoid Obstacles Using Reinforcement Learning for Mobile Robots

Uses Deep Deterministic Policy Gradient (DDPG) based reinforcement learning to develop a strategy for a mobile robot to avoid obstacles. For a brief summary of the DDPG algorithm, see Deep Deterministic Policy Gradient Agents.

Open Live Script