The ability of a humanoid robot to stand up again after a fall is fundamental for autonomy. In a scenatio like RoboCup, getting up can be a challenge. On the one hand, the robot must be ready to play ready to play again as quickly as possible, but there are also many factors that can prevent the robot from getting up and fast standing up. These include internal factors, such as worn joints, joint play, or overheated joints, and external factors, such as collision with other players or even "getting stuck" on the artificial turf.
If it is detected that the robot has fallen down, then a stand-up movement is performed. The current solution is based on a static KeyFrame approach. A series of predefined poses (motions) are played, which lead to the robot poses (movements) are played, which lead to a high probability that the robot will stand up afterwards. This static sequence of motions does not adapt to the situation. Deviations that occur during standing up can lead to the robot falling down again. fall down. This includes being pushed by other robots or getting stuck on the lawn.
The task is to analyze the current solution for its problems and develop a new dynamic variant with sensory feedback. For example, the position sensors and the joint sensors could be used to check sensors could be used to check whether the expected poses have been achieved. If necessary, corrections can be made. Related solutions from other teams will be used as a reference, e.g. B-Human. The developed solution should then be empirically evaluated and compared with the old solution.
Machine learning methods can also be used to optimize the motion.
Other possible aspects to consider:
If the fall is detected at an early stage, the arms and head can be moved in such a way that the active fall becomes safer and standing up can happen more quickly. and getting up can happen more quickly.
Relevant publications and theses:
Possible procedure for the final thesis
We imagine the process as follows:
- Discussing the topic and getting to know each other
- Writing an exposé
- Discussion of the exposé and possible adjustments
- Registration of the work
- Performing experiments in the simulator and on the robot if necessary.
- Evaluating the experiments
- Writing the thesis
- Writing a short summary of the results for the NaoTH team report.
Independently send a short status to the Nao team every two weeks.
Information about Nao Team Humboldt
RoboCup Regeln: https://cdn.robocup.org/spl/wp/2021/01/SPL-Rules-2021.pdf
Public Github: https://github.com/BerlinUnited
Internal Gitlab: https://scm.cms.hu-berlin.de/berlinunited
Source code of other RoboCup SPL teams
Public NaoTH Repo: https://github.com/BerlinUnited/NaoTH
Public Nao Devils Repo: https://github.com/NaoDevils/CodeRelease
Public B-human Repo: https://github.com/bhuman/BHumanCodeRelease
Bembelbots: https://github.com/Bembelbots rUNSWift: https://github.com/UNSWComputing/rUNSWift-2019-release
Publication lists of other teams
Possibly work of other teams can be helpful:
B-Human Abschlussarbeiten: https://b-human.de/theses-en.html
B-Human Publikationen: https://b-human.de/publications-en.html
Reviewer: V. Hafner
Adaptive Systems: https://hu.berlin/adapt
Supervision and technical support: Heinrich Mellmann, Stella Alice Schlotter.
Lab work required: yes
Own PC required: yes