Description:Provisional
The idea is to use sensor data from mobile phones to identify the faulty road conditions. Every vehicle
on road is like a “mobile phone on wheels”. As the vehicle goes over faulty road condition, the jerks
and bumps are also felt by the mobile phone in the vehicle. We collect the data in real time from mobile
phones to identify the vertical linear acceleration due to bad roads. Using unsupervised learning AI
models, the exact location and the severity of the defect can be identified in consideration of other
parameters like speed, two dimensional horizontal linear acceleration.
Complete
All vehicles traverse in three dimensional space. The horizontal 2 dimensional plane is used for
movement of vehicle on the road. The vertical dimension comes into action only when the surface is
uneven which will happen in one of the cases of potholes, broken patches and undulations.
The basic idea is to capture vertical linear acceleration of vehicle with respect to horizontal linear
acceleration and velocity in horizontal two dimensional space. Considering every vehicle is a “mobile
phone on wheels” and every mobile phone comes with builtin accelerometer as part of basic sensor kit.
The accelerometer captures acceleration across three dimensions. As the vehicle moves in straight line
the acceleration and retardation will be across one dimension plane only. When the vehicle goes
through a turn, acceleration/retardation will be across second dimension as well, creating a two
dimensional plane. The third vertical dimensional impact will be very minimal and can be ignored for
smooth roads. If the vehicle goes over undulation or broken patch or potholes or speed breaker, it will
result into movement across third dimensional (vertical) plane, the third dimension will show
movement. The movement of any vehicle on smooth road is in XY plane where as movement on broken road will
have a component along Z axis in addition to XY plane. The pattern of movement along Z axis can
differentiate between pothole and speedbreakers.
Capturing the movement across third dimensional plane along with GPS latitude and longitude
coordinates enables the system to pinpoint the exact location and severity of the road defect. To
eliminate the bias of inverse proportionality between speed and vertical movement, the speed of
moving vehicle is also captured to correctly calibrate the severity of the road defect. The data points are fed to the unsupervised learning AI model which translates them to a centroid along
with the severity of the defect represented by different color pins.
Tackling problem of difference in sensor sensitivity calibration - Different mobile phones use different
sensors based on cost, quality and accuracy. These sensors have different calibration sensitivity which
includes accelerometers. The problem is further compounded by the differently calibrated GPS. As a
result, the same defect is reported by different mobiles at slightly different points and different severity
levels. The problem is solved by unsupervised learning AI models. A custom built algorithmic is used
to find the centroid and builds a heat map around it based on different data points. The defects on the system appear in almost real time. The system also allows features like geo tagging
and associate different sensitivity to different geo tagged regions.

There are two broad personas
being defined with their respective user interfaces.
1. The end user view - This view is available on the mobile phone in a form of a layer on the navigation
application (google maps, MMI, tom tom, etc) which shows the road defects as the user is driving
along (represented by blue dot). The severity of defects is shown by pins of different colors, red being
critical, amber as high and green as low. The system also provides an option of audio alerts about
approaching a critical road defect. Here no action or input is needed by end user. All the pins appear via
system inteligence.
2. The authority view - This view enables the authority to view road networks of a city/district/sate.
The view enables the authority to view defects in real time. This view also enables the history of the
defect for example when it was created and how the severity of the defect is changing over a period of
time. This view enables authority to close a defect only when its fixed, it will remove the
corresponding pins from the user view and authority view. In case the fix is not proper, the defect will
reappear on its own with latest details, as the latitude and longitude still remain the same, this new
defect will be added to the history of previous view.
, Claims:Claims
1. Safer and smooth roads - The system will bring in instant reporting system making roads safer for
traffic and avoid accidents due to bad roads. It will also enable the respective authorities in real time
about exact location and severity of road defects.
2. Mental peace of driving: Reduced cases of road rage
3. Helping environment: reduction in CO2 by eliminating one of the cause of traffic jams and
enabling vehicles to run in their efficiency band.
4. Scientific way of quantifying quality of roads without any human intervention. This also
eliminates the biases.
5. Warns driver of upcoming conditions, hence giving them a lead time to take corrective actions.
6. Huge Cost saving : Saves huge cost for the authorities, time and effort of physical inspection of the
road to identify defects . Save on service and car maintenance because of the better road conditions