Introduction
Welcome to the blog of the ITU ZES Solar Car Team Autonomous Systems Group! In
this blog, we aim to share our journey, projects, and experiences in the exciting field of
autonomous vehicles.
About Us
Our Mission
Our mission is to advance autonomous vehicle technology by transforming our main solar car into a fully autonomous vehicle. We aim to compete in the World Solar Challenge (WSC), anticipating future regulations that may require teams to incorporate Advanced DriverAssistance Systems (ADAS) and autonomous driving features. Our goal is to develop an SAE Level 4 autonomous car capable of safely navigating real-time traffic conditions. While building the car and its autonomous systems, we are committed to testing our progress through participation in various competitions, ensuring that our solutions are robust and ready for real-world applications.
Autonomous Systems Group
As a autonomous systems group of ITU ZES Solar Car Team, we are composed of passionate students from various engineering disciplines at ITU. Our members are mainly pursuing majors in:
Robotic and Autonomous Systems Engineering
Computer Engineering
Electronics and Communications Engineering
AI and Data Engineering
We expect other engineering majors with a keen interest in autonomous systems into our group.
Each member brings a unique set of skills and perspectives, contributing to the multidisciplinary nature of our autonomous vehicle projects. We have organized our team into specialized sub-teams to focus on key areas of development. Our collaborative environment fosters innovation and learning, allowing us to tackle complex challenges in autonomous vehicle technology. Together, we are committed to pushing the boundaries of what’s possible and making significant contributions to the field.
What We Do
Our team is organized into specialized sub-teams, each focusing on key aspects of autonomous vehicle technology:
Perception and Sensing
The Perception sub-team is responsible for enabling the vehicle to interpret and understand its surroundings using sensors and advanced computer vision algorithms.
2-D Object Detection Models for Signs and Traffic Lights:
We develop and train 2-D object detection models to accurately identify road signs and traffic lights. This involves creating custom datasets specific to our operating environments and employing state-of-the-art deep learning techniques to ensure high detection accuracy in various lighting and weather conditions.
Creating Datasets and Training Models:
Building reliable perception systems requires robust datasets. Our team collects and annotates data to create comprehensive datasets that include images of traffic signs and lights. We use this data to train our object detection models, continuously refining them to improve performance.
LiDAR Point Cloud Data Filtering:
Raw LiDAR data can be noisy and contain irrelevant information. We utilized filtering techniques to clean and preprocess point cloud data, enhancing the quality of the data used for further processing.
Clustering Algorithms for Point Cloud Data:
To detect and interpret objects in the environment, we apply clustering algorithms to the filtered point cloud data. This helps in segmenting the data into meaningful clusters representing potential obstacles or landmarks.
3-D Object Detection Models:
Beyond 2-D detection, we are working on 3-D object detection models that leverage LiDAR and camera data to identify objects in three dimensions. This enhances the vehicle’s understanding of its surroundings, allowing for more precise navigation and obstacle avoidance.
LiDAR and Camera Fusion:
Sensor fusion is a critical component of our perception system. By combining data from LiDAR sensors and cameras, we create a more comprehensive and accurate representation of the environment. This fusion enhances object detection and classification capabilities, improving overall system reliability.
Through these efforts, our Perception sub-team aims to create a robust and reliable perception system that enables the autonomous vehicle to navigate safely and efficiently in complex real-world environments.
Localization and Mapping
The Localization and Mapping sub-team is dedicated to accurately determining the vehicle’s position within its environment and creating detailed maps for navigation. Our work involves integrating data from multiple sensors and employing advanced algorithms to achieve precise
and reliable localization.
Sensor Fusion:
We utilize a combination of sensors, including LiDAR, Inertial Measurement Units (IMU), and Global Navigation Satellite Systems (GNSS), to collect comprehensive environmental and positional data. By fusing data from these sensors, we enhance the accuracy and robustness of the vehicle’s localization capabilities.
Mapping Algorithms:
Our team employs state-of-the-art Simultaneous Localization and Mapping (SLAM) algorithms to construct high-resolution maps of the environment. These algorithms enable the vehicle to map its surroundings in real-time while simultaneously keeping track of its own position within the map.
Localization Techniques:
For precise localization, we utilized methods such as Extended Kalman Filters (EKF) and scan matching techniques. These approaches refine the vehicle’s position estimates by comparing incoming sensor data with existing map information.
Global to Local Position Conversion:
We convert global positioning data into local coordinates within our maps. This transformation allows the vehicle to understand its position relative to the mapped environment, facilitating accurate navigation and route planning.
Artificial Lane Generation:
To guide the vehicle along desired paths, we design artificial lanes within our maps. These lanes serve as reference trajectories for the vehicle’s planning and control systems, ensuring smooth and efficient movement through various environments
By integrating these methods, our Localization and Mapping sub-team provides a foundational layer that enables the autonomous vehicle to navigate safely and effectively. Our focus on sensor fusion and advanced algorithms ensures that the vehicle maintains accurate positioning, which is critical for autonomous operation in real-world conditions.
Simulation and User Interface
The Simulation and User Interface sub-team plays a crucial role in developing and testing our autonomous vehicle systems.
Simulation:
Simulation allows us to test and validate our algorithms in a virtual environment before deploying them on the actual vehicle. By simulating real-world driving scenarios, we can identify and address issues in a safe and controlled manner. We use simulation programs such as Gazebo and CARLA to replicate vehicle dynamics and environmental conditions. These tools help us refine our perception, localization, planning, and control algorithms without restricted by the physical conditions.
User Interface (UI):
Our user interface is a web application designed to simplify the operation of the autonomous vehicle systems. It is used for starting the programs and subsystems of the vehicle, as well as monitoring feedback from the car. The UI allows team members to control various modules and observe real-time sensor data and system performance, facilitating efficient testing and development.
Planning and Control
The Planning and Control sub-team is responsible for the vehicle’s decision-making processes, including path planning, obstacle avoidance, and motion control. Our work ensures that the autonomous vehicle navigates safely and efficiently by utilizing various algorithms and planning strategies.
Planning Algorithms
Our planning system consists of multiple layers to generate optimal paths and behaviors:
Global Planner: Determines the overall route from the starting point to the destination using algorithms like Dijkstra’s and A*. This planner takes into account the map data to find the most efficient path.
Mission Planner: Handles mid-level planning tasks, such as setting waypoints and adjusting routes based on dynamic conditions.
Behavioral Planner: Decides how the vehicle should behave in response to the environment and traffic rules. It makes decisions like lane changes, overtaking, and yielding.
Our planning algorithms integrate data from all vehicle systems—such as perception and localization—to create plans that adapt to different driving situations. This layered approach allows the vehicle to make informed decisions at both strategic and tactical levels, ensuring
safe and effective navigation in real-time conditions.
Control Algorithms
We implement control algorithms to manage the vehicle’s motion, ensuring stability and accurate path following. These algorithms help the vehicle maintain desired speeds and steering angles while adapting to changing conditions.
The control algorithms we use include:
Pure Pursuit
Proportional-Integral-Derivative (PID)
Stanley Method
Linear Quadratic Regulator (LQR)
Model Predictive Control (MPC)
Competitions
We actively participate in several competitions to benchmark our technologies and collaborate with the broader autonomous vehicle community.
Teknofest (2021-2024)
Description: Teknofest is an annual aerospace and technology festival held in Turkey. It features a wide range of technology competitions, exhibitions, and events aimed at promoting innovation among the youth. Our team participated in the autonomous vehicle competitions at Teknofest in 2021, 2022, 2023, and 2024.
Website: www.teknofest.org/tr/competitions/competition/29
Shell Eco-Marathon Autonomous Programming Competition (2023)
Description: The Shell Eco-marathon Autonomous Programming Competition is an international event that challenges students to design, program, and test autonomous vehicles with a focus on energy efficiency and innovative engineering solutions.
Website: www.shell.com/energy-and-innovation/shell-ecomarathon.html
Bosch Future Mobility Challange (2023)
Description: The Bosch Future Mobility Challenge is an international competition
organized by Bosch Engineering Center Cluj. It invites university teams to develop autonomous driving algorithms and implement them on scale-model vehicles to navigate complex urban environments.
Website: www.boschfuturemobility.com
Alumni Feedback
İsmet Atabay (Formal Group Leader - 2022):
It is inevitable that autonomous vehicles, which are currently under development, will
become a part of our lives in the future. In addition to solar-powered vehicles, ITU
Solar Car Team started this journey in 2019 in order to both show itself and develop
its vision in this future technology and has shown itself in international competitions.
In this process, taking on the autonomous group responsibility for 3 years has shown
me the dedication, working order, winning and producing technology in this field while
I was still a student. Thanks to being a part of this team, it has enabled me to continue the same work I did in the team when I graduated from the team in professional
business life. Therefore, I would like to thank all alumni and active team members,
my colleagues and advisors for bringing this team from 2004 to these days.
Muhammet Yavuz Köseoğlu (Formal Group Member - 2022):
One of my most valuable experiences during my university life was being a member
of ITU SCT. Not only did I have the opportunity to develop my own interests, but I
also made lifelong friendships. This team is not only a technical or academic platform,
but also a unique environment where you can develop your skills such as teamwork,
creativity and leadership. Founded in 2004, the team offers a significant contribution to
your future career by enabling you to have contacts from numerous and diverse sectors.
One of the most important things I learned when I joined the team was how powerful
an impact working together and supporting each other can have. If you want to spend
your university years more meaningful and full, ITU SCT is the place for that. This is
a great opportunity to sign new projects, get to know and develop yourself. Be a part
of this family, you will not regret it!
Serkan Aksöz (Formal Group Member - 2022):
The years I spent at ITU Solar Car Team were the most enjoyable times of my
university life. Like every engineer, I was theoretically improving myself and learning
things in my spare time, but as a student you don’t have the opportunity to get your
hands dirty and do something. I applied to the team thanks to a friend I met in an
online course during pandemic, we built a vehicle from scratch and made it autonomous
for the next 2 years, we shed a lot of blood, sweat and tears in this process, we made
very good friends, we still keep in touch. Both our technical and social skills have
improved a lot. Thanks to what we learned in this process, many of us had internships
and jobs before we graduated, I recommend everyone to first be a student at ITU and
then be a part of the Solar Car Team.
Talha Işık (Formal Group Member - 2022):
My joining the autonomous system group of ITU SCT started unexpectedly when
one of my friends said, “I found such a team, I’m going, do you want to come?”.
The three of us had already shaped this newly established autonomous system group
together. After joining the team, we formed strong friendships with both the other
groups and the new friends we recruited, and we still keep in touch. I still remember
our late night and even early morning work and tests together. The offer I received
from where I work now was also thanks to the success we achieved in that team. I
continue to work in the field of robotics with great pleasure and passion.
Kadir Aksu (Formal Group Member - 2023):
ITU ZES Solar Car Team was not only a place where I developed my technical knowledge and skills, but also an environment where I learned how to apply this knowledge
in real life and opened important doors for me into the business world. Every technical detail, competition experience and the unity and solidarity within the team gave
me a great advantage in my transition to professional life. The strong friendships we
established in this process, where I learned team spirit and the culture of working together, have been one of the most valuable contributions of the team. Thanks to these
friendships established in the team, a wide network has been formed for more than 20
years from all over the world. I would like to thank all the friends we worked with in
this process and everyone who supported us.
Javad İbrahimli (Formal Group Member - 2023):
Taking part in the Autonomous Systems Group of ITU ZES Solar Car Team was one
of the most important steps in my engineering career. Contributing to the development
of autonomous driving systems allowed me to apply my advanced knowledge of software,
hardware and algorithms in the real world. Taking part in these projects made me
realize the importance of teamwork and interdisciplinary collaboration. In addition,
creating innovative and sustainable technology-based solutions has contributed greatly
to both my professional and personal development. My most important advice for new
members of the team is to approach the projects with full curiosity and determination,
because this team will add vision to you beyond everything you learn.”
Comments