Scribd
Upload
7 views4 pages

My Presentation in Details

Uploaded by

haseeb29581
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
7 views4 pages

My Presentation in Details

Uploaded by

haseeb29581
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Slide #02

1. Objective: Implement a deep learning-based control system for automated


vehicles.
The aim is to create a smart control system for self-driving cars using deep learning.
This system will help the car make driving decisions on its own, like a human driver,
and handle different driving situations effectively.
2. Challenge: Maneuvering in environments without GPS or external
communication.
The main challenge is to make the car drive safely in places where GPS signals are
weak or unavailable, like tunnels or remote areas. Additionally, the car won't rely on
communication with other vehicles or servers. It needs to use only the sensors on
board to understand its location, avoid obstacles, and make driving decisions.
3. Approach: Use of sensor fusion, SLAM (Simultaneous Localization and
Mapping), and Model Predictive Control (MPC).
 Sensor Fusion: This means combining data from different sensors (like
cameras and INS) to better understand what's around it. It makes the
information more accurate by using data from different sensors..
 SLAM (Simultaneous Localization and Mapping): This technique allows
the car to create a map of its environment while figuring out its own location
on that map. It helps the car navigate in places where no pre-made maps are
available.
 Model Predictive Control (MPC): This is a way for the car to plan its moves
ahead of time. It predicts future positions and chooses the best actions to
take, like changing lanes or slowing down to avoid obstacles, ensuring
smooth and safe driving.

Slide #03

Tools and Techniques


1. Pygame: We use Pygame to create pictures and animations for our project. It
helps us show how our vehicle moves on the screen.
2. Kalman Filter: This tool helps us clean up the sensor data. It removes any
errors or noise from the readings, making it easier to understand where the
vehicle is.
3. SVM Model: SVM stands for Support Vector Machine. This model helps us
predict what the vehicle should do based on the data we have. It helps decide
the best actions for the vehicle.
4. SLAM: SLAM means Simultaneous Localization and Mapping. This method
helps the vehicle find out where it is on a map while also creating that map.
It’s important for the vehicle to know its location to move correctly.
5. MPC: MPC stands for Model Predictive Control. This tool helps us make good
decisions about how the vehicle should act. It looks ahead to see what might
happen and tells the vehicle what to do to get the best results.
Environment
In our project, the vehicles drive on roads without using GPS or any communication.
They rely on their own sensors and the tools mentioned above to find their way and
avoid obstacles.

Slide #04

Sensor Data
1. INS (Inertial Navigation System): This system collects real-time data from
three types of sensors:
o Accelerometer: Measures how fast the vehicle is speeding up or
slowing down.
o Gyroscope: Detects the vehicle's orientation and rotation.
o Magnetometer: Helps determine the vehicle's direction by sensing the
Earth's magnetic field.
2. Simulated Camera: Detects objects around the car by drawing a circle to
show its range. It highlights any nearby vehicles and can adjust how much it
can see (field of view). You can zoom in by pressing "Q" or zoom out by
pressing "A."
Vehicle Behavior
The ego vehicle can perform various actions:
 Lane Changing:
o Space in adjacent lane (120 meter)
 Overtaking:
o After lane change check space for overtake (200 meter ahead)
 Following:
o Follow when no space in adjacent lane for performing lane change
maneuver and when reach the distance (80 meter) then speed same to
the closest vehicle ahead
 Keeping Lane: Staying within the lines of its lane.
 Stopping:
o Stop when reach the distance (80 meter) and closest vehicle is stop.
The vehicle makes decisions in real-time by analyzing the positions of nearby
vehicles and the information from its sensors.

Slide #06

What is SLAM?
SLAM is a system that helps a vehicle figure out where it is and create a map of its
surroundings at the same time. Here’s how it works:
1. Kalman Filter:
o It helps keep track of the vehicle’s position (where it is) and speed
(how fast it’s moving).
o It also finds important things on the road, like signs or other landmarks,
and marks them on the map.
2. Mapping and Visualization:
o The landmarks that the vehicle finds are shown as green circles on
the map. This helps the vehicle know where these important spots are.
o The system makes sure not to show the same landmark twice, which
keeps the map clear and easy to understand.

Slide #07

What is Model Predictive Control (MPC)?


MPC is a system that helps a vehicle plan its movements and stay on the right path.
Here's how it works:
1. Prediction Horizon:
o It predicts where the vehicle will be in the future based on what it’s
doing now and what movement it’s planning next.
o This helps the vehicle plan ahead and make smoother maneuvers.
2. Cost Function:
o The goal is to keep the vehicle as close to the best path as possible.
o It also makes sure the vehicle stays a safe distance away from other
cars on the road.
3. Maneuver Optimization:
o The system chooses the best action, like changing lanes, following
another car, or passing a slower vehicle.
o It adjusts these actions on the go to make driving smoother and safer.

You might also like