You are on page 1of 1

mm-Wave Radar and Machine Learning-Based Parking Places Monitoring

Summary:
In this project, an mm-wave FMCW radar will be used to update the state of parking locations. For this
project, Infineon’s Distance2Go module will be used, which is one transmit and one receive antenna
FMCW radar module. Its field of view is 20x80 and can detect metal targets from 25 meters. FMCW
signal will be transmitted and received. Applying Fourier transform on the received signal ranges of
different targets can be found, if a vehicle is present the reflected signal will be from the vehicle if not it
will be from the ground. If the vehicle is not present, the reflected signal from a particular parking
location will be from the ground, since the ground is not a metasurface, the received signal will be weak
and show slightly more range corresponding to the corresponding parking location. Therefore, only
finding the range will not be enough. To tackle this problem, machine learning will be exploited here.
Machine learning will train a model with all possible scenarios and decide on the new input data based
on this model. A low complexity machine learning model will also be explored that will depend only on
the individual locations.

Initially, we will use a five locations parking place and will monitor it with a single radar. The system will
map all five locations on the application display, whenever there is a vacant place it will be shown on the
display.

Benefits:
In contrast to the available systems, where to monitor each parking place we require a dedicated
sensor, the proposed system can monitor multiple vehicles. Therefore, the system cost can be
significantly reduced. Moreover, the proposed system is applicable in all weather conditions.

Prerequisite: 
Basic knowledge of FMCW radar, Python, Matlab, or C++.

You might also like