Demo: Carla Simulation of automotive Dead Recking Based-on Error State Kalman Filter

In this demo, a Kalman filter is applied to track the pose of an automotive. IMU data and speed information¹ are collected through Carla rosbridge. In the update step, the speed measurement is used to update the states. Moreover, non-holonomic constraints are applied.

Prerequisite

• Linux and ROS (has been test on Ubuntu 20.04 with ROS Noetic).
• Carla Simulator and Carla rosbridge. Make sure your Python can find carla package, Python 3.7 is recommended.

Build

• First, clone into you catkin workspace

cd catkin_ws/src   
git clone https://github.com/LHengyi/ESKF-for-Automotive-Dead-Reckoning.git   

• Build

catkin_make  

• To run
  start Carla server

CarlaUE4.sh

roslaunch automotive_dead_reckoning carla_localization_ad_rosbridge.launch  

in another terminal

roslaunch automotive_dead_reckoning wheel_ins.launch  

Demonstration

Footnotes

1. In practice, vehicle speed information can be obtained through on-board interface, visual odometry or even by binding an IMU on a non-steering wheel. ↩