Projektpraktikum RoboCup@Home
Practical course for the TUM Winter/Summer Semesters
Goals of the practical course:
We aim to form a team of students for the RoboCup@Home competition, which is the largest competition for service robots. The focus of this competition is on challenging scientific problems such as human-robot interaction, mobile manipulation, and cognition. Then, the student team needs to design, implement and test perception, learning, and robot control algorithms for these challenges. The overall goal is to provide students with enough practical background to participate in the international annual competition of the RoboCup@Home league. The service robot needs to demonstrate its abilities through the following functionalities:
- Navigation
- Mapping of known or unknown environment
- Person recognition
- Person tracking Object recognition
- Object manipulation
- Speech recognition
- Gesture recognition
- Cognition to understand current situations.
Then, in order to address the above topics, this practical course is split as follows:
Kick-off meeting: The students meet with a supervisor to form groups according to the students' knowledge. The students designate a team leader (preferably a master student) who stays in touch with one of the supervisors of this course.
Hands-on tutorials: The students get introductory tutorials regarding methods, software, and equipment used for this course. These tutorials will assure that all students have the same general knowledge of the used methods. Development and test phase: The students will design and implement algorithms to solve the above tests. This development goes hand in hand with the validation/test phase.
Final phase: Final test scenarios on a mobile robot are evaluated using one scenario from a previous competition to measure the performance of the robots abilities.
This practical project aims to select those students who are motivated enough to continue working on developing algorithms to address the tests of the competition throughout the year. This practical course should contribute to the long-term goal of establishing a RoboCup@Home student team to participate in the official competition. For example, the competition in 2017 took place in Japan.
Learning and Cognition
Navigation and Environment Mapping
Object Manipulation and planning control (Pictures courtesy of PAL robotics taken during the RoboCup@Home competition 2016 in Leipzig. )
Then, in order to address the above topics, this practical course is split as follows:
- Kick-off meeting: The students meet with a supervisor to form groups according to the students' knowledge. The students designate a team leader (preferably a master student) who stays in touch with one of the supervisors of this course.
- Hands-on tutorials: The students get introductory tutorials regarding methods, software, and equipment used for this course. These tutorials will assure that all students have the same general knowledge of the used methods. Development and test phase: The students will design and implement algorithms to solve the above tests. This development goes hand in hand with the validation/test phase.
- Final phase: Final test scenarios on a mobile robot are evaluated using one scenario from a previous competition to measure the performance of the robots abilities. This practical project aims to select those students who are motivated enough to continue working on developing algorithms to address the tests of the competition throughout the year.
This practical course should contribute to the long-term goal of establishing a RoboCup@Home student team to participate in the official competition. For example, the next competition will be in 2017 and it will be in Japan.
Practical course schedule
This practical course is given two times per week in the following schedule:
- Monday 15:00 - 16:30 hrs
- Wednesday 15:00 - 16:30 hrs
Learning outcome
Upon successful competition of this practical course, students are able to:
Implement algorithms from different areas including perception, learning, and control in the context of service robots. Evaluate these algorithms in simulation and on a real mobile robot.
Analyze and apply techniques in challenging and real scenarios such as homes or supermarkets.
Besides the technical skills learned in this practical course, the students are able to organize and manage a large engineering project.
Furthermore, the students will learn communication skills to explain the results achieved per groups.
Requirements for the lecture
On the theoretical level, bachelor students must have passed their first two semesters. It is highly recommended to have basic knowledge in control, vision, or learning methods. On the practical level, students must have: Strong C++ programming skills. Background on ROS is highly advised. Prior basic knowledge on AI (PROLOG) and Robotics is highly recommended.
Evaluation of the practical course
This practical course is evaluated with laboratory assignments and a final group project. Each introductory tutorial session is composed of one or more exercises and students need to solve them individually. Then, students are split into groups and each group needs to implement algorithms to solve different problems according to the competition requirements in order to enable different functionalities to a mobile robot. The teams' final projects are evaluated with a presentation and live demonstrations of the implemented algorithms to summarize their accomplishments. In addition to the technical implementations, each team needs to deliver a written report which should reflect the ability to analyze and understand scientific and technical problems related to service robots.
The final grade is composed as follows:
Final group project, including presentation and final report : 70%
Individual laboratory assignments : 30%
Suggested literature
We recommend the students the following literature:
- RoboCup@Home: Analysis and results of evolving competitions for domestic and service robots, Iocchi L., Holz D., Ruiz-del-Solar J. , Sugiura K. and van der Zant T. Artificial Intelligence, Volume 229, December 2015, Pages 258–281, DOI: dx.doi.org/10.1016/j.artint.2015.08.002
- Artificial Intelligence: A modern approach. Stuart Russell and Peter Norvig. Pearson Ed. Springer Handbook of Robotics. Bruno Siciliano and Oussama Khatib. 2007. Springer-Verlag New York, Inc., Secaucus, NJ, USA.
- Machine Learning. Tom Mitchell. McGrawHill Robot modeling and control. M.W. Spong, S. Hutchinson, and M. Vidyasagar. John Wiley & Sons, 2006.