This post discusses how to develop a low cost treadmill with a closed-loop feedback controller for reinforcement learning experiments.

MATLAB and JAVA code is linked.

Code & Links

• Pre-build java library
• Source code of the developer version

The Treadmill

• Get a standard household treadmill Samples
• Note: It should work with a DC-Motor, otherwise a different controller is needed!

The Controller and the Distance Sensor

• Pololu Jrk 21v3 USB Motor Controller with Feedback or stronger (max. 28V, 3A)
• Comes with a Windows Gui to specify the control gains
• Sharp distance sensor GP2Y0A21, 10 cm – 80 cm or similar
• USB cable
• Cable for the distance sensor
• Power cables for the treadmill
• Controller User Guide by Polo

The Matlab Interface

• Get the java library  build or the developer version, both from Sept 2015 created by E. Rueckert.
• Run the install script installFTSensor.m (which add the jar to your classpath.txt)
• Check the testFTSensor.m script which builds on the wrapper class MatlabFTCL5040Sensor (you need to add this file to your path)

LEGO EV3 for Robotic Tasks

We have five EV3 sets and use them for studying robot control, motion planning and visual navigation from depth images. 

We use our GitHup LEGO Python project for our developments. 

Tactile Sensing

Several special purpose sensors including depth image cameras (shown in the center in the image), IMUs, accelerometers, gyroscopes, sonic sensors (two are shown in the image), etc. can be connected to the EV3 brick. 

The EV3 systems can be used to explore neural sensor fusion approaches, embedded computing implementations and classical mobile robotics tasks.  

• Details to our LEGO Python GitHub project

Videos

https://vimeo.com/501651310https://vimeo.com/374166607

Human-inspired, Adult-size Dexterous Robot Hand

We use a adult-sized robot hand for learning grasping and object manipulation skills. The hand is mounted on our FRANKA EMIKA Panda robot. 

The hand has 19 degrees-of-freedom and uses 8 smart actuators for precise control (actuators contained inside the unit).

Under actuated design aims to provide the right balance between fine control and conformance to the shape of the objects.

Tactile Sensing

All actuators provide real time control and feedback of position, speed and current measurement (with direction), enabling inference of applied force.

Additional data including actuator temperature, (over)load status and PWM, a Palm ToF Distance sensor and optional Capacitive pads at the back of the palm complete the sensor array.

We also have five 3-axis force-torque sensors (FTS) (shown in the image) attached to each finger tip. The FTS measure contact force and shear forces with a resolution of 1mN / 0.1g.

• E-Manual of the RH8 hand.
• ROS Interface description of the RH8 hand.
• E-Manual of the 3Axis F/T Sensors 
• Details to our FRANKA ROS GitHub project.
• Details on the robot hardware can be found on the SEED Robotics page.

Videos

• Research videos using the robot will be presented here. 

Publications

2020

Xue, H.; Boettger, S.; Rottmann, N.; Pandya, H.; Bruder, R.; Neumann, G.; Schweikard, A.; Rueckert, E. Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks Proceedings Article In: International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020), 2020. Links | BibTeX @inproceedings{Xue2020, title = {Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks}, author = {H. Xue and S. Boettger and N. Rottmann and H. Pandya and R. Bruder and G. Neumann and A. Schweikard and E. Rueckert}, url = {https://cps.unileoben.ac.at/wp/ASPAI2020Xue.pdf, Article File}, year = {2020}, date = {2020-06-30}, booktitle = {International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020)}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Close Article File Close

We are developing a repository for real-time control of the FRANKA EMIKA Panda 7-dof robot arm.

Our project is based on ROS and allows to teleoperate the robot arm in real-time using motion tracking data provided by OptiTrack’s Motive software. 

GitHub Project and Links

• https://github.com/ai-lab-science/Franka-ROS-TRAIN
• Details on the robot hardware
• Or on the developer site on world.franka.de
• ROS Gazebo Tutorial

Videos

• Research videos using the robot will be presented here. 

Publications

2020

Xue, H.; Boettger, S.; Rottmann, N.; Pandya, H.; Bruder, R.; Neumann, G.; Schweikard, A.; Rueckert, E. Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks Proceedings Article In: International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020), 2020. Links | BibTeX @inproceedings{Xue2020, title = {Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks}, author = {H. Xue and S. Boettger and N. Rottmann and H. Pandya and R. Bruder and G. Neumann and A. Schweikard and E. Rueckert}, url = {https://cps.unileoben.ac.at/wp/ASPAI2020Xue.pdf, Article File}, year = {2020}, date = {2020-06-30}, booktitle = {International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020)}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Close Article File Close

FRANKA EMIKA’s Panda robot arm is a complient, light-weight robot arm with seven degrees-of-freedom. 

We use the C++ libfranka library in our own ROS project for learning complex manipulation skills. 

Links

• Details to our FRANKA ROS GitHub project
• Details on the robot hardware can be found on the FRANKA EMIKA page
• Or on the developer site on world.franka.de

Videos

• Research videos using the robot will be presented here. 

Publications

2020

Xue, H.; Boettger, S.; Rottmann, N.; Pandya, H.; Bruder, R.; Neumann, G.; Schweikard, A.; Rueckert, E. Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks Proceedings Article In: International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020), 2020. Links | BibTeX @inproceedings{Xue2020, title = {Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks}, author = {H. Xue and S. Boettger and N. Rottmann and H. Pandya and R. Bruder and G. Neumann and A. Schweikard and E. Rueckert}, url = {https://cps.unileoben.ac.at/wp/ASPAI2020Xue.pdf, Article File}, year = {2020}, date = {2020-06-30}, booktitle = {International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2020)}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Close Article File Close

Nils Rottmann, M.Sc. has developed a tutorial on using ROS and Gazebo. 

This tutorial was used in our humanoid robotics and machine learning lectures. 

GitHub Code & Links

• https://github.com/NRottmann/ROS_Gazebo_Tutorial

Weitere Links und Tutorials

• Python Videolectures
• Python Online Tutorials

Dieses open-source Projekt enthält Tools und Demos für die Python-Entwicklung mit den Lego Mindstorms EV3 und EV3Dev Bricks. 

Die Inhalte sind verständlich aufbereitet und wir haben zahlreiche Tutorials und Aufgaben für Schüler*innen erstellt. 

GitHub Code & Links

• https://github.com/ai-lab-science/LEGORoboticsPython
• Dieses Repository ist Teil unseres Lego Robotik Schüler*innen Projekts.  

Details to the Software Development

Dieser Einführungsvortrag beschreibt die grundlegenden Schritte um einen LEGO Roboter zu bauen und mit Python zu programmieren. 

Weitere Links und Tutorials

• Wiki: https://github.com/rueckert/LEGORoboticsPython/wiki
• Installation und Dokumentation von Micropython.

Sensor gloves are gaining importance in tracking hand and finger movements in virtual reality applications as well as in scientific research. In this project, we developed  a low-budget, yet accurate sensor glove system that uses flex sensors for fast and efficient motion tracking. 

The contributions are ROS Interfaces, simulation models as well as motion modeling approaches. 

GitHub Code & Links

• https://github.com/ai-lab-science/SensorGloves
• The software development is part of the TRAIN project. 
• The development is based on prior sensor glove designs.

Details to the Software Development

The figure shows a simplified schematic diagram of the system architecture for our sensor glove design:

(a) Glove layout with sensor placements, the orange fields denote the flex sensors, while the IMU is marked as a green rectangle,

(b) Circuit board which is wired with the sensor glove, has 10 voltage dividers for reading each flex sensor connected to ADC pins of the microcontoller ESP32-S2 and the IMU is connected to I2C pins,

(c) The ESP32-S2 sends the raw data via WiFi as ROS messages to the computer, which allows a real-time visualization in Unity or Gazebo,

(d) Post-processing of the recorded data, e.g. learning probabilistic movement models and searching for similarities.

Publications

A research publication by Robin Denz, Rabia Demirci, M. Ege Cansev, Adna Bliek, Philipp Beckerle, Elmar Rueckert and Nils Rottmann is currently under review.