TeeTUM

Inspiration

The rapid advancement of autonomous vehicle technology presents both opportunities and challenges in urban transportation. Efficiently managing a fleet of robotaxis requires sophisticated algorithms to optimize assignments while considering factors like customer wait times and environmental impact. We were inspired to create TeeTUM to explore these challenges and contribute to the development of sustainable and efficient autonomous transportation systems.

What it does

TeeTUM simulates a robotaxi fleet management system that assigns vehicles to customers using various algorithms, including greedy, k-nearest neighbors, and the Hungarian algorithm. The platform collects and analyzes metrics such as average wait times, ride durations, total distance traveled, and vehicle utilization rates. By comparing the performance of different assignment algorithms on identical scenarios, TeeTUM provides insights into the most effective strategies for fleet management.

How we built it

We built TeeTUM using FastAPI for the backend, creating RESTful endpoints to interact with a scenario runner that simulates the robotaxi environment. We implemented several assignment algorithms and integrated them into an asynchronous assignment loop. For data analysis, we collected metrics during simulations and used tools like Pandas and Matplotlib to visualize the results. The codebase is modular, with clearly defined models, routers, and algorithm modules, facilitating scalability and future enhancements.

Challenges we ran into

One of the main challenges was ensuring accurate state synchronization between the simulation environment and our application, especially when tracking vehicle and customer states during assignments. Managing asynchronous execution and time tracking within the simulation loop required careful handling to prevent race conditions and ensure data integrity. Implementing multiple algorithms and comparing their performance also involved optimizing computational efficiency without compromising the accuracy of the simulation.

Accomplishments that we're proud of

We’re proud of successfully implementing multiple assignment algorithms and integrating them into a functional simulation platform. Our ability to collect and analyze performance metrics allowed us to gain valuable insights into the efficiency of different algorithms under various conditions. Overcoming technical challenges related to asynchronous programming and state management has resulted in a robust and scalable application that can serve as a foundation for further research and development.

What we learned

Through this project, we deepened our understanding of algorithmic optimization, asynchronous programming, and real-time data handling. We learned the importance of accurate state management in simulations and the complexities involved in balancing multiple objectives like customer satisfaction and environmental impact. Additionally, we gained experience in using FastAPI for building APIs and in data analysis techniques for interpreting simulation results, which are valuable skills for future projects.

What's next for TeeTUM

In the future, we plan to enhance TeeTUM by incorporating more advanced algorithms, such as machine learning-based predictive models for demand forecasting. We aim to develop a user-friendly frontend interface to allow interactive simulations and real-time visualization of metrics. Integrating additional constraints like vehicle capacity, time windows, and dynamic pricing models will make the simulation more realistic and applicable to real-world scenarios. Ultimately, we hope TeeTUM can contribute to the development of efficient and sustainable urban transportation systems.

Built With

• angular.js
• fastapi
• python