Skip to content

Gaussian Process localization with ToF and RSSI

Notifications You must be signed in to change notification settings

Miikkasna/gploc

Repository files navigation

Gaussian Process for ToF and RSSI localization

This repository holds Gaussian Process implementation for a localization task. The task is a miniature car racing localization. The data is collected using OpenAI gym car racing simulator. The data was scaled to match 30m x 30m training plane. Two datasets were then generated from the data; Time of Flight (ToF) data and Received Signal Strength Indication (RSSI) data. ToF data was sampled at 1Hz and RSSI data with 5Hz. Noise in data was Gaussian Distributed with zero mean and standard deviation of 0.15m for ToF and 1.0m for RSSI. Error in speed measurement is 0.3m/s for both. See the blog text for more information.

The structure of the dataset is following:

real_position.npy : 4 features, [time (s), speed (m/s), x-position (m), y-position (m)]

tof_1hz.npy : 4 features, [time (s), speed (m/s), x-position (m), y-position (m)]

rssi_1hz.npy : 4 features, [time (s), speed (m/s), x-position (m), y-position (m)]

Data visualization alt text GP predicted route visualization alt text GP predicted real-time visualization

alt text

About

Gaussian Process localization with ToF and RSSI

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published