From Strategy to Victory: SZEnergy Team’s Winning Formula at Shell Eco-marathon Europe

From Strategy to Victory SZEnergy Team's Winning Formula at Shell Eco-marathon Europe

Join the SZEnergy team as they share their inspiring journey from 6th place to record-breaking Mileage Challenge winners, revealing their unwavering dedication and MATLAB optimization. Discover how innovation and engineering prowess powered their victory in the Autonomous Challenge.

  • 1833

The Department of Automotive and Railway Engineering at Széchenyi István University created the SZEnergy Team with the primary objective of advancing the efficiency of electric-driven vehicles and proving the viability of their cutting-edge technology. Commencing its journey in 2008, the Team has actively participated in the world’s largest energy efficiency challenge, the Shell Eco-marathon (SEM), in the “Urban Concept” category. Notably, in both 2022 and 2023, the SZEnergy Team established a new world record, secured victory in the Battery-Electric category of the SEM Mileage Challenge, and finished at the top of the podium at the SEM Autonomous Challenge 2023.

This incredible success wouldn’t have been possible without our dedicated Team members, but it’s also thanks to our smart choice of software. MATLAB and Simulink have been essential in helping us reach our goals.

Shell Eco-Marathon

The Shell Eco-Marathon (SEM) is a preeminent and highly esteemed competition in the realm of energy efficiency competitions, with a longstanding tradition in Europe dating back to 1985. The contest is designed for engineering students worldwide and functions as a platform for innovative methodologies in developing fuel-efficient automobiles.

The SEM Mileage Challenge

The SEM Mileage Challenge invites students to design, build, and race energy-efficient vehicles, which must run approximately 16 kilometers in under 40 minutes.

Urban Concept

This competition allows participants to choose from two vehicle classes: “Prototype” and “Urban Concept” (UC). The main goal of the UC category is to replicate urban transportation scenarios, including slow speeds and frequent stops.

The SEM Urban Concept Autonomous competition

The Autonomous category contest was first introduced in 2019, taking the SEM competition to a whole new level. As a part of it, teams are required to consider the application of autonomous technology to their Urban Concept vehicle. The vehicle must operate fully autonomously and without any intervention from the safety pilot.

The Challenge consists of three main sections: autonomous track drive, obstacle avoidance, and parking maneuverability. The final score is calculated based on the team’s on-track performance and a business presentation describing their approach to the competition and vehicle design.

SZEnergy Team - Winner of Autonomous and Mileage Challenge
Image 1: SZEnergy Team – Winner of Autonomous and Mileage Challenge

The SZEnergy Team and the project

Our team has been participating in the Urban Concept category of SEM since 2008. Initially, our vehicles were powered by solar energy, but starting from 2013, all of our SEM race vehicles have been constructed for the Battery Electric category. Our most recent race car, named SZEmission, made its debut in 2019.

The team, along with our sponsors, worked together to design and build this vehicle from scratch. The advanced carbon monocoque chassis and use of optimized parts made from aluminum resulted in an optimized vehicle mass of 95 kg. The self-designed power and system electronics consume below 2W during standby operation. Low electric consumption while in motion has been made possible from a self-designed motor controller and a unique BLDC motor.

The vehicle had its inaugural race at the 2019 SEM event in London, where our team achieved an impressive 6th place in the Mileage Challenge. Unfortunately, the physical SEM competition didn’t take place in 2020 and 2021 due to the COVID-19 pandemic. This unexpected break allowed us to dedicate more time to vehicle development and fine-tuning the vehicle’s characteristics in a MATLAB environment.

Our debut in autonomous racing took place in 2022, and since then, we have been continuously improving the entire system, focusing on both hardware and software aspects.

An autonomous system consists of three main components: main sensors, actuators, and an on-board computer (OBC).

The main sensors are 128-channel LIDAR and a Dual Stereo Camera; additionally, there are a high-precision inertial measurement unit and wheel encoders for localization.

The OBC consists of two high-performance embedded computers, which are responsible for conducting the main computational tasks, and there is a real-time computer with an FPGA for providing direct controls for the actuators. Of the two OBCs, one is designated to manage all the necessary camera stream processing, while the other carries out the planning and path-tracking processes.

SZEmission - on the track
Image 2: SZEmission – on the track

The Role of MATLAB and Simulink software

We developed a longitudinal mathematical vehicle model using MATLAB and Simulink environments. This model allowed us to incorporate various factors, including the measured resistance force model, powertrain characteristics, and track conditions, into sub-assemblies.

To obtain accurate model parameters, we conducted field tests and performed test-bench measurements on the electric drivetrain of the vehicle. The reference torque for the motor controller served as the input for our model. We optimized this torque input using MATLAB’s built-in genetic algorithm (GA).

During the optimization process, the GA leveraged the Simulink model in each iteration to calculate the precise electric consumption of the vehicle while completing the race distance. The resulting optimized torque reference vector provided the driving strategy that our driver followed during the Mileage Challenge.

This optimization environment, supported by MATLAB tools, played a pivotal role in our having achieved 291 km/kWh at the Circuit Paul Armagnac in Nogaro as the winner of the race.

Image 3: Overview of the vehicle model © Pusztai, Z.; Kőrös, P.; Szauter, F.; Friedler, F. Vehicle Model-Based Driving Strategy Optimization for Lightweight Vehicle. Energies 2022, 15, 3631. https://doi.org/10.3390/en15103631
Image 3: Overview of the vehicle model © Pusztai, Z.; Kőrös, P.; Szauter, F.; Friedler, F. Vehicle Model-Based Driving Strategy Optimization for Lightweight Vehicle. Energies 2022, 15, 3631. https://doi.org/10.3390/en15103631

Our Team adopted a unified approach to tackle all the challenges, utilizing a single-stack autonomy strategy. This means that we didn’t develop specialized solutions for each track section. Instead, our focus was on creating a comprehensive representation of the environment using an occupancy grid map, which contains all the essential information required by the motion planner, encompassing details about the drivable road surface and obstacles.

To obtain precise depth information, we fused the data from the LIDAR Point Cloud with the Camera Images stream. Achieving accuracy required knowing the exact translation and rotation parameters between the LIDAR sensor and the camera. For this purpose, we found that the MATLAB LIDAR Camera Calibrator App was the most effective tool.

With the assistance of MATLAB, we achieved a high level of accuracy by analyzing 20 pairs of point cloud images, allowing us to obtain precise parameters. Subsequently, we integrated these parameters into the ROS Transformation Server, enabling us to apply these transformations across all the nodes in our system.

Within our Team, we rely on a robot operating system (ROS) for all communication and deployment within the autonomous system. During the calibration process, we diligently compiled all sensor data in a ROSBAG file. To extract and utilize this data in the Camera Lidar Calibrator App, we developed a small MATLAB script that relies on the ROSBAG extractor.

Gathering different samples with the calibration target
Image 4: Gathering different samples with the calibration target

To accurately follow the given trajectory, our autonomous stack incorporated a system for providing smooth steering inputs to the actuators. We implemented a path-tracking algorithm using the MATLAB and Simulink environments. Since our entire autonomous system was built on ROS middleware, we needed MATLAB and Simulink to be able to communicate with existing ROS Nodes.

Thanks to the ROS toolbox, facilitating communication between the trajectory planning and path tracking nodes was relatively easy. Within the Simulink environment, we easily integrated the Pure Pursuit path-tracking algorithm, benefiting from MATLAB’s pre-implemented function.

To deploy the ROS nodes onto the vehicle, we had to convert them into C++ code, and we accomplished this using the Simulink Coder App. This allowed us to utilize the C/C++ code on the vehicle itself or on any other computer. During the deployment process, we encountered some challenges with optimizing the code within the MATLAB environment. As a result, we opted to use the default C++ compilers for code optimization.

Ensuring robust control was paramount in the Autonomous Challenge, as it enabled us to maintain smooth and precise vehicle control across all sections of the track.

The Simulink Code for the Path Tracking Algorithm
Image 5: The Simulink Code for the Path Tracking Algorithm

 

Conclusion

Winning the Shell Eco-Marathon requires more than just a well-designed and well-built car. Understanding the vehicle’s operation and effectively modeling it are essential for success. MATLAB played a crucial role in our journey by providing a powerful optimization environment that transformed our car from a 6th-place finisher into a record-breaking Mileage Challenge-winning vehicle, all with minor hardware modifications.

Securing victory in the autonomous challenge was a significant milestone for our team. It fulfilled a longstanding aspiration, and we recognized the potential for impressive results in this category with focused efforts. Moving forward, we remain committed to enhancing our autonomous system and are convinced MATLAB is a valuable tool for supporting our ongoing improvements and innovations.

Featured products

MathWorks® products:

 

Learn more about SZEnergy Team

Recommended Events

Recommended Posts

Excitation Signals for Identification of Dynamic Systems

Excitation Signals for Identification of Dynamic Systems

Learn about optimizing multisine signals for precise frequency-domain system identification. Discover the importance of low Crest Factor (CF) in excitation signals and explore techniques to enhance signal-to-noise ratios (SNR) for improved measurement accuracy.

Ai robotics working on a car

AI: Driving the Industry Towards Greater Success

Artificial intelligence (AI) is seen as a promising technology that can help leading OEMs to maintain their position as market leaders. In this post, read about how AI is changing the manufacturing sector, as well as its potential advantages and potential drawbacks.

MATLAB and Simulink for Automotive

MATLAB and Simulink for Automotive

Discover how MATLAB and Simulink drive automotive innovation. Learn how these tools expedite vehicle development and help OEMs meet evolving market demands.