Sprungmarken

Servicenavigation

Hauptnavigation

Sie sind hier:

Hauptinhalt

Detailseite



F1/10 Autonomous Race Car

Chair: Regelungssystemtechnik & Institute for Informatics 12

Supervisior: Frank Hoffmann, Mikail Yala,

Start: 07.10.2019

Maximum amount of participants: 12

Description: Tracing back the history of autonomous cars, experiments for autonomous (self-driving) cars have been conducted since at least the 1920s. This vision has gradually become reality and is subject to serious world-wide interest. Along with the technical maturation of electric cars, autonomous and electric cars are getting increasingly market-ready.

Towards this trend, using 1/10th-scale miniature RC race cars (RC-car), e.g., the Traxxas F1 type platform, with augmented sensors and computing hardware as a platform to demonstrate autonomous driving and SLAM (Simultaneous Localization and Mapping) concepts have started to emerge recently in graduate-level researches and courses.

The F1/10 student competition f1tenth.org provides a venue for designing control and perception algorithms on an open source hardware and software (ROS) platform.

This project is supervised in collaboration by the chair RST at IRF and the institute of informatics. It is designated to spark the student's interests towards autonomous mobile robotics and real-time systems as well as the founding of a competitive F1/10 Autonomous Racing team from TU Dortmund to participate in the international F1/10 Autonomous Racing Competition in the following year.

Both chairs already assembled the platform and the necessary components, e.g., competitive Traxxas F1 type platforms, sensors, racing tracks and an experimental environment. Based on this, students have to design, build and test their autonomous navigation strategy, real-time controls (model predictive control), reinforcement learning on a 1/10th-scale mini race car. Moreover, students will learn to use the robot operating system (ROS), integrate various sensors (IMU, Cameras, LIDAR) on an embedded computer (Nvidia Jetson TX2) and implement their own algorithms for localization, mapping, path planning, and control. At the end of this project, there will be a competition to determine the best team in TU Dortmund in a "Time Trials" competition.

We provide some checkpoints and guidelines to help the students get started. Here are the specific tasks including:
-Demonstrate navigation strategies via the Gazebo-based simulator provided by F1/10 organization.
-Make the car drive autonomously over the whole lap without any manual control.
-Participate in the final competition with proposed algorithms.
-Further design algorithms for aggressive maneuvering (model predictive control), online trajectory planning, and overtaking strategies.
- System identificaton of vehicle dynamic parameters and an advanced vehicle model (slippage, lateral acceleration limits)
- Tuning of motor and velocity controllers

Students are expected to have profound knowledge in programming, basic knowledge in statistics and probability, basic knowledge in Linux. Programming experience in Matlab, ROS, or C++ is helpful. Previous attendance of the course Mobile Robots is helpful.


 

The project group takes place!

Participants: 215848, 215130, 215802, 214604, 214227, 181236, 214956,