Localization algorithm

---

• Localization algorithm
  • Install
  • Data Description
    • Rosbag
    • CSV files
  • Technical description
  • Performance analysis
  • Work description

---

In this repository, I developed a localization algorithm for a car equipped with wheel encoders and a global navigation satellite system (GNSS) receiver. The wheel encoders provide linear and angular speed information, while the GNSS provides position information.

The following assumptions are made:

• The sensors are mounted in the center of the rear axle of the car.
• The car is moving on a 2D plane.
• The position is expressed in the map frame. The map and odom frames are identical in our scenario.
• The measurements are affected by some noise, the estimated covariances of which are provided.

The expected output of the algorithm is the following:

• Estimated position and orientation of the car in the map frame.

Install

The code comes with a ready-to-use conda environment:

git clone https://github.com/Davidelanz/localization-algorithm
cd localization-algorithm
conda env create -f environment.yml 
conda activate localization-algorithm
jupyter-lab

With the previous commands, you should be able to navigate easily the notebooks provided in the repository at http://localhost:8888/lab.

Data Description

The input is provided both as a rosbag and CSV files.

Rosbag

The rosbag data/data.bag contains the following topics:

• /sensors/gnss/odom: GNSS position measurements of the car in the map frame.
• /sensors/odom: linear and angular speed measurements of the car with respect to the odom frame, in the frame of the car (x axis pointing forward, y axis pointing left)

CSV files

For all CSV files, the time is given as a Unix timestamp (in nanoseconds), position is in metre, and speed in metre/second. The provided files are the following:

• data/gnss.csv: GNSS 2D position measurements of the car in the map frame. The estimated standard deviation on the measurements is 0.7 m.
• data/odom.csv: linear and angular speed measurements of the car with respect to the odom frame, in the frame of the car (x axis pointing forward, y axis pointing left). The covariances on the measurements are also provided.
• data/ground_truth.csv: true position and speed of the car, given for reference. The orientation is expressed as a quaternion. The speed is given with respect to the odom frame, in the frame of the car.

Technical description

Technical description of the work

Performance analysis

Performance analysis of the algorithm

Work description

The work description along with the various development phases is made available in the work_description.ipynb notebook.