Help Roboy Let's build our future.

Hire Roboy

Support the project by booking Roboy for your event.
We would be thrilled to attend your company party, christmas party or any other occasion with Roboy. He can be programmed to say what you would like or simply entertain your guests with Roboy’s presence – make your event unforgettable by having a remarkable guest.

For any enquiry send an e-mail to gro.y1498689402obor@1498689402sseni1498689402sub1498689402!

Help Roboy

Upcoming Events

 

25.08.2017 | Roboy Demo Night

UnternehmerTUM, Lichtenbergstraße 6, 85748 Garching bei München

Traditionally in the end of each semester Roboy Team is hosting a demo night where we show what we’ve been working on for the past half a year.

21.10.2017 – 22.10.2017 | Hack Roboy #3

UnternehmerTUM, Lichtenbergstraße 6, 85748 Garching bei München

After the great success of Roboy’s second Hackathon in April 2017, we are ready for a new one!

Become a Partner

While we don‘t pay our students, building Roboy is still expensive.
To make our project a reality we need your support: monetarily, or with your products & technology.
There is a share of Roboy for you!

Help Roboy
 
 
 

Products

Implementing your products or tools in Roboy can be exciting showcase for your company’s portfolio.

Technology

Integrating your software into Roboy can show new applications, and possibly expand your portfolio into different markets.

Know-How

We treat the Roboy team as a small business and need to constantly educate ourselves. Your valuable expertise in form of consulting, workshops, trainings or implementation ideas can help us reach our goals.

Money

Your monetary support gives us the opportunity to cover the additional materials, hardware, and production parts that have not already been provided by a sponsor.

Become a Team Member

Benefits

  • Apply your theoretical knowledge to a hands-on project
  • Join an interdisciplinary and international team of students
  • Learn from supervising PhDs, experts, workshops and invited speakers
  • Get access to the new MakerSpace workshop at UnternehmerTUM
  • Get the chance to travel with Roboy to fairs, conferences and international representative event
 
Help Roboy

We Want You

  • You are highly interested in robotics
  • You are a good team player
  • You can offer valuable skills to the team
  • You are excited to learn new things
  • You are a student at a University (or similar) in Munich – PhD students, post-doc and professors are also welcome!

If this is you, we invite you to join our team of enthusiastic and smart creatives! You can get involved in one or several divisions: robotics, programming, fundraising, marketing, design…

Or you inspire us with your excellent ideas!

Get in touch by emailing gro.y1498689402obor@1498689402maet1498689402.

Thesis

Want to help the project? Want to work with us? Here’s a list of open topics!

If you are interested in pursuing one of the projects, drop us a line at gro.y1498689402obor@1498689402maet1498689402.

We are also open to your ideas, let us know!


Neuralnetwork-based Control of an Anthropomimetic Robot

Project Description

Background

Service robots “living” in close interaction with humans are
a prospective future technology with a wide field of applications. Anthropomimetic robots, built according to the morphology of the human body, are a favorite branch of development as they may better integrate into humanrobot
ecologies. Moreover, such human-like constructs, controlled by neural
networks, may also improve our understanding human behavior and development.
World wide, several of such muscle tendon driven systems have
already been built. While mechatronically at a very advanced level, the control of such systems is still in its infancy. The thesis will study and further develop a novel form of neural networks which have especially been developed for controlling such “soft” embodied systems. As recent experiments have confirmed, these brain-body systems develop spontaneously individual behaviors in the interaction with physical objects or humans, see the videos bottleswing, bottleshakehorizontal, and bottleshakevertical. A further example for the behavior spectrum is MyoArm-Widemoves The great advantage of the method is that the robot is kind of creative but also open to learning its behaviors by direct physical interaction with humans, see for instance MyoArm-handshake-small. This partly circumvents the demanding problem of programming the behavior and should be of immediate interest for using such systems in future robot technology.

Description

These networks are applied to various systems, preferentially using
a Myorobotics arm – a generic muscle-tendon driven system. Theses are storage and recall of behavior primitives for building higher level behavior architectures (Georg Martius) and (ii) methodical developments and inclusion of vision in order to study the emergence of visuomotor coordination (Ralf Der). Eventually, the neurocontroller will be applied to the Roboy platform to generate whole-body motion patterns in soft interaction with humans.

Requirements

Level Master Thesis
Areas Software Engineering & Control
Required Skills C++
Available Institutions TUM

 

Documents

No documents at this time.

Project Progress

%


Gaming Roboy

Project Description

thesis-teaser

Background

Roboy is above all a communications project. One of the main goals is to communicate about robotics and moral implications thereof to a wider public as well as to motivate pupils to study in the STEM subjects. As games are one of the primal means of communications to reach today’s tech-savvy youth, we are looking for a team of skilled programmers that develop games with and around roboy.

Description

There is no definite concept of how the game might work. It can be a serious game that explains moral issues, or one that teaches about relevant robotics (e.g. Kerbal Space Program for Robots). Or it can be a pure fun game, like a Jump & Run where Roboy is the main character. Or, we have a complete script for a childrens’ book, how about making it a interactive story?  Alternatively, Roboy could be the playground for a “microbot race”. Or why not build a game where Roboy is part of? Remember the childrens game, where you’re only allowed to walk towards the opponent when he’s not looking, but have to stand still when he looks at you? Or maybe track and augment Roboy to create a new game experience?

We’re open to your suggestions, we provide resources where needed and a thriving community that will be eager to help and beta test your ideas.

Requirements

Level Master Games Praxis
Areas Game Engineering
Required Skills Unity, C#, C++
Available Institutions TUM

 

Documents

No documents at this time.

Project Progress

%


Dynamic Muscle Unit

Project Description

thesis-teaser

Background

To build robots with muscle-like actuation an imitation for muscles has to be found. There are many different approaches in use robotics, from pneumatic Mc Kibben muscles, electro active polymers, shape memory alloys, contracting coiled fishing line through temperature changes to electro motors coiling up a tendon. Each of these approaches has its own specific up- and downsides.
In the Myorobotics EU-FP7[1] project, we are building a muscoloskeletal modular robotics toolkit based on the last principle with electro motors.[2] The two most important advantages are that thanks to tendon routing we can move the muscle’s mass away from  the extremities, building light and soft and therefore safe robots. And as electro motors are safe and easy in handling and there are no large external parts like a compressor required, the final product may be operated by non-specialist users and is easy to transport.
The motor units developed in the Myorobotics unit are being  developed from scratch and while they show great performance, they are still relatively expensive, limiting dissemination of the concept. We are therefore striving to create a second flavor of  muscle that is building on top of the Dynamixel motor units. These inexpensive units have sensing and communication embedded only requiring a system to coil up a tendon and a non-linear spring.

Description

The task of this project is to design, build, model and calibrate a cost-effective addon to the Dynamixel motor unit that implements the two core concepts required for muscular control: Coiling of the tendon and non-linear elasticity. The technology to be used are off the shelf parts combined with laser-sintered free-form parts exlusively. The unit will be designed as a SolidWorks part and once finished it is then to be measured and modeled in our robotics simulation and control framework MYODE.

Requirements

Level Master Thesis
Areas Robotics, Mechanical Engineering, Electrical Engineering
Required Skills SolidWorks, C++
Available Institutions TUM

References

[1] “Myorobotics modular muscoloskeletal robotics toolkit,” http://myorobotics.eu.
[2] Myorobotics: A modular toolkit for musculoskeletal robot development, Hong Kong,
China, Jun 2014.
[3] M. Jäntsch, S. Wittmeier, K. Dalamagkidis, A. Panos, F. Volkart, and A. Knoll, “Anthrob – a printed anthropomimetic robot,” in Proc. IEEE-RAS International Conference on Humanoid Robots (Humanoids), 2013.

Documents

No documents at this time.

Project Progress

0%


Closing the Simulation Gap

Project Description

thesis-teaser

Background

Roboy[1] is a muscoloskeletal robotic demonstration platform that showcases the use of soft interaction with humans in a complete humanoid system. While its soft actuation makes it safe to interact with and extremely robust, precise control and planing is hard. In order to master the complexity for learning these complex dynamics a good simulation model is important. By building on simulation techniques developed at our chair, the existing roboy CAD model and the Gazebo Robotics simulation toolkit a simulation of Roboy will be created and evaluated. This will lay the ground-work to use Roboy for simulation experiments with the Brain Simulation Platform currently in development in the Human Brain Project.

Description

The task of this project is to build a simulation model of Roboy in the  Gazebo Robotics Simulation Framework. Thereby complex problems e.g. arising from tendon wrapping and friction need to be solved.

Requirements

Level Master Thesis
Areas Robotics, Software Engineering, Rigid Body Simulation
Required Skills SolidWorks, C++, Gazebo
Available Institutions TUM

References

[1] R. Pfeifer, P. Y. Tao, H. Gravato Marques, S. Weydert, D. Brum, M. Weyland, R. Hostettler, F. Volkert, V. Gmünder, and D. Halbeisen, “Roboy anthropomimetic robot,” 2013. [Online]. Available: http://www.roboy.org
[2] M. Jäntsch, S. Wittmeier, K. Dalamagkidis, A. Panos, F. Volkart, and A. Knoll, “Anthrob – a printed anthropomimetic robot,” in Proc. IEEE-RAS International Conference on Humanoid Robots (Humanoids), 2013.
[3] M. Jäntsch, “Non-linear control strategies for musculoskeletal robots,” Ph.D. dissertation, Universitätsbibliothek der TU München, 2014.
[4] “ECCEROBOT embodied cognition in a compliantly engineered robot,” http://eccerobot.org.
[5] “Myorobotics modular muscoloskeletal robotics toolkit,” http://myorobotics.eu.
[6] Myorobotics: A modular toolkit for musculoskeletal robot development, Hong Kong, China, Jun 2014.

Documents

No documents at this time.

Project Progress

0%


Knowledge Based Dialog Management

Project Description

thesis-teaser

Background

Roboy[1] is a muscoloskeletal robotic demonstration platform that showcases the use of soft interaction with humans in a complete  humanoid system. Thanks to its soft actuation it is safe to interact with and extremely robust, making it the ideal testbed for social human-robotics-interaction. Furthermore, due to regular public events, new interactions can be tested on a regular basis. However, to generate this interaction the robot needs to be able to reason about its current state and the state of interaction. In this project we make use of a semantic descriptional language implemented in the Web Ontology Language (OWL)[2] to describe all aspects of tasks[3] which shall be carried out by Roboy. OWL is part of the Semantic Web initiative of the World Wide Web Consortium and is based upon a strong formal specification, thus allowing for automatic reasoning about the represented knowledge. A natural language interface, including speech recognition, speech synthesis and the management of dialogues shall be created and connected with the rich semantic descriptions. This enables the system to better ground the referents of utterances and to generate adequate responses to the human operator’s requests.

Description

The task of this project is to design and implement a dialog and robot state management system based on preexisting open-source software modules available as ROS packages in the Gazebo  simulation software. Furthermore knowledge gained in the James project can be reused. For this project to succeed a clear interface specification between the knowledge base, perception and dialog system is paramount. Finally, the system design should make it easy to integrate new tasks, e.g. to achieve active sensor guidance.

Requirements

Level Master Thesis
Areas Robotics, Computational Linguistics, Knowledge
Representation, Natural Language Processing
Required Skills ROS, OpenCCG, Solid
programming skills (Java, Python, C++), Gazebo
Available Institutions TUM, Fortiss

References

[1] R. Pfeifer, P. Y. Tao, H. Gravato Marques, S. Weydert, D. Brum, M. Weyland, R. Hostettler, F. Volkert, V. Gmünder, and D. Halbeisen, “Roboy anthropomimetic robot,” 2013. [Online]. Available:  http://www.roboy.org
[2] S. Rudolph, B. Parsia, P. Hitzler, M. Krötzsch, and P. Patel-Schneider, “OWL 2 web ontology language primer (second edition),” W3C, W3C Recommendation, Dec. 2012, http://www.w3.org/TR/2012/REC-owl2-primer-20121211/.
[3] M. Tenorth, A. Perzylo, R. Lafrenz, and M. Beetz, “Representation and Exchange of Knowledge about Actions, Objects, and  Environments in the RoboEarth Framework,” IEEE Transactions on Automation Science and Engineering (T-ASE), 2013.
[4] C. Bozsahin, G.-J. M. Kruijff, and M. White, “Specifying Grammars for OpenCCG: A Rough Guide.” [Online]. Available:  http://www.utexas.edu/courses/jbaldrid/2006/computational-syntax/papers/openccg_guide.pdf

Documents

No documents at this time.

Project Progress

0%


MuscleARoboy - AR based muscle visualisation of a muscoloskeletal robot

Project Description

Background

Roboy is a muscoloskeletal robotic research platform. That is, it mimics the human body’s musculoskeletal system through tendons attached to motor units that behave like simplified muscles. One of the future applications for Roboy is a trainer for prospective doctors and physiotherapists. While the motor units display comparable dynamics to muscles, they look very different. Furthermore many larger muscles don’t have singular attachment points but larger attachment areas or multiple heads. For doctors to understand how the muscular interplay creates movement, it is important to see the muscles in a realistic fashion. With the increasing availability of augmented reality – either through devices such as mobile phones and tablets or through head mounted displays – a logical approach to this problem is to augment the camera image of roboy with a  realistically simulated muscle depiction. With such a system, the doctor can physically interact with the robot and at the same time
get an intuitive idea, how this interaction translates into changes in muscle length.

Description

This is project consists of several tasks, that can be solved by  different students independently or in a group:

Vision Design

a system that recognizes the posture of the robot as well as the field of view of the observing camera and renders a figure on top of the camera image, based on the robots posture.

Pose Estimation

Design a system that estimates the pose of the robot, based on all the available sensory data of the robot.

Muscle Fitting

Design an algorithm that takes the pose of the robot as an input and matches a muscoloskeletal system to the given pose.

Requirements

Level Master Thesis
Areas Robotics, Computer Vision, Augmented Reality, Computer Graphics
Required Skills Solid Programming Skills (C++, Java, …)
Available Institutions TUM

References

[1] G. Klein and D. Murray, “Parallel tracking and mapping for small AR workspaces,” Mixed and Augmented Reality, 2007. ISMAR 2007. 6th IEEE and ACM International Symposium on, pp. 225–234, 2007.
[2] V. Pradeep, C. Rhemann, S. Izadi, C. Zach, M. Bleyer, and S. Bathiche, “MonoFusion: Real-time 3D reconstruction of small scenes with a single web camera,” pp. 83–88, 2013.
[3] R. A. Newcombe and A. J. Davison, “Live dense reconstruction with a single moving camera,” Computer Vision and Pattern Recognition (CVPR), 2010 IEEE Conference on, pp. 2606–2613, 2010.
[4] R. A. Newcombe, S. J. Lovegrove, and A. J. Davison, “DTAM: Dense Tracking and Mapping in Real-Time,” ICCV ’11: Proceedings of the 2011 International Conference on Computer Vision, pp. 1–8, Aug. 2011.
[5] B. Ummenhofer and T. Brox, “Dense 3D Reconstruction with a Hand-held Camera,” in Pattern Recognition (Proc. DAGM), Springer, 2012.

Documents

No documents at this time.

Project Progress

0%


MyoAnthrob - Anthropomimetic Robotic Arm

Project Description

thesis-teaser

Background

In the ECCEROBOT[1] Project the robotic arm Anthrob has been built that features muscles in order to imitate the control paradigms that govern human motion. In the successor project Myorobotics[2], a modular toolkit for this kind of muscoloskeletal robots has been developed. The next step is to build a Myorobotics-based modular arm with the same configuration as Anthrob to show the improvements of Myorobotics over the previous solution and have a biomimetically actuated system available for research.

Description

The task of this position is to design and build a Myorobotics shoulder joint primitive that can be connected to the current primitives as well as a setup of the full arm with eleven muscle modules.

Requirements

Level Master Thesis
Areas Robotics, Software Engineering, Rigid Body Simulation
Required Skills SolidWorks, C++, Gazebo
Available Institutions TUM

 References

[1] “ECCEROBOT embodied cognition in a compliantly engineered robot.” http://eccerobot.org.
[2] “Myorobotics modular muscoloskeletal robotics toolkit.” http://myorobotics.eu.
[3] M. Jäntsch, S. Wittmeier, K. Dalamagkidis, A. Panos, F. Volkart, and A. Knoll, “Anthrob – a printed anthropomimetic robot,” in Proc. IEEE-RAS International Conference on Humanoid Robots (Humanoids), 2013.
[4] M. Jäntsch, Non-Linear Control Strategies for Musculoskeletal Robots. PhD thesis, Universitätsbibliothek der TU München, 2014.
[5] Myorobotics: A modular toolkit for musculoskeletal robot development, (Hong Kong, China), Jun 2014.

Documents

No documents at this time.

Project Progress

%


Roboy Interaction Framework

Project Description

thesis-teaser

Background

The regular public displays of Roboy make it the ideal platform to test different theories of human-robotics-interaction (HRI), with a diverse group of people meeting Roboy for the first time. However, up until now, this potential is unused as there is no conclusive framework to design, implement and evaluate HRI experiments on Roboy.

Description

The goal of this project is to develop a framework that allows to quickly implement and test new ideas in HRI. To achieve this, an interface together with a set of building blocks is to be created. These building blocks will make it easy for the end user to control Roboy and to have him react based on rules and states to situations. Furthermore, the framework should make it easy to gather  Experiment data and evaluate it according to the research question.

Requirements

Level Master Thesis
Areas Robotics, Human-Robotics-Interaction, Software Engineering, Psychology
Required Skills C++, C#
Available Institutions TUM

Documents

No documents at this time.

Project Progress

0%


Robotic Multi Material Printing

Project Description

thesis-teaser

Background

In the last few years layered additive manufacturing – commonly called 3d printing – has seen a surge of new applications and implementations. Mainly, because the original patents have run out and enthusiasts have started building their own printers, thereby founding a grass root movement that is gaining traction throughout many industries. Especially fused filament fabrication (FFF) printers are available at very affordable prices, orders of magnitude below the market prices for professional machines. Lately, stereo lithography machines have entered the hobbyist market as well and first attempts at selective laser melting are currently pursued. However, each of these printing technologies has their own advantages and disadvantages and one carefully has to choose manufacturing technology based on the needs of each part. Furthermore it is not possible to combine the printing technologies with other, traditional fabrication methods such as machining.
We suggest a solution built around a 7-DOF robot arm readily  available at our laboratory.
This arm is carrying the build platform and brings the product in the making to different printing heads mounted within reach. This allows for the combination of different materials as well as manufacturing technologies.

Description

The task of this project is to set up the robot arm and two material disposition heads; an ordinary fused filament printing head, as well as a head that dispenses a tendon. This will allow to embed the tendon inside the printed object, be it to create movable joints,  stronger materials or tensegrity structures. In the second part of the project, a software has to be developed that generates the trajectories for the arm as well as the printing head.

Requirements

Level Master Thesis
Areas Rapid Prototyping, Robotics
Required Skills Mechanical & Electrical Engineering, Software development
Available Institutions TUM

Documents

No documents at this time.

Project Progress

0%


SkinnyBoy - Angle reconstruction on a soft robot, through a soft skin

Project Description

thesis-teaser

Background

Proprioceptive measurements of the 3d angle of a ball-in-socket joint in the humanoid robot Roboy[1] is a yet to be solved problem. While a reconstruction from the measured tendon lengths is possible, the values are imprecise due to friction and the resulting hystereses. In a recently completed project a sensorized surface has been developed[2] which allows for 3d reconstruction[4] of its configuration. Through the application of this technique and by
fixing it to the two rigid bodies attached to the joint, we hope to recover the joint angle.[3]

Description

The task of this project is to implement the reconstructive algorithms in Roboy’s framework and then to apply the reconstruction technique to estimate the current ball-in-socket joint angle.

Requirements

Level Master Thesis
Areas Robotics, Electrical Engineering, Programming
Required Skills Good knowledge of linear
algebra, sensor fusion, good
programming skills (C++)
Available Institutions TUM

References

[1] R. Pfeifer, P. Y. Tao, H. Gravato Marques, S. Weydert, D. Brum, M. Weyland, R. Hostettler, F. Volkert, V. Gmünder, and D. Halbeisen, “Roboy anthropomimetic robot,” 2013. [Online]. Available:  http://www.roboy.org
[2] P. Mittendorfer and G. Cheng, “Humanoid multimodal tactile-sensing modules,” Robotics, IEEE Transactions on, vol. 27, no. 3, pp. 401–410, 2011.
[3] ——, “Open-loop self-calibration of articulated robots with artificial skins,” in Robotics and Automation (ICRA), 2012 IEEE International Conference on. IEEE, 2012, pp. 4539–4545.
[4] ——, “3d surface reconstruction for robotic body parts with artificial skins,” in Intelligent Robots and Systems (IROS), 2012 IEEE/RSJ International Conference on. IEEE, 2012, pp. 4505–4510.

Documents

No documents at this time.

Project Progress

%


Tendon Force Sensor for 3D Printed Robots

Project Description

thesis-teaser

Background

When building musculoskeletal robots, it is important to know the force that is applied at the muscle attachments. To date, there are no good solutions to measure this force. Currently, in the two toolkits that are being developed – Roboy and Myorobotics – the force measurement is integrated in the motor unit and measured through displacement of a linear spring, that also provides the muscles elasticity. This solution, while having many advantages, does not account for the force lost due to friction when the cable is routed. As friction can be quite significant and is notoriously hard to model, measuring the applied force after all routing points is very important to control the robot.

Description

The task of this project is to design, build, model and calibrate a cost-effective sensor that measures the force magnitude at the end of the tendon based on ideas developed at the chair. The technology to be used are off the shelf parts combined with laser-sintered free-form parts. The sensor will be designed as a SolidWorks part that can be embedded into larger 3D printed parts such as bones or clamps. The finished sensor will then be measured and modeled in our robotics simulation and control framework MYODE.

Requirements

Level Master Thesis
Areas Mechanical Engineering
Required Skills SolidWorks, C++
Available Institutions TUM, ZHAW

 

Documents

No documents at this time.

Project Progress

%