Real-time Control of Actuated Systems
Module II
Coursename Einführung in die Echtzeit-Kontrolle von aktuierten Systemen
Course number in TuCan 03-41-0582-se
Type Seminar with practical tutorials
Credits 5 CP
Offered every summer semester

This module is aiming at teaching students

  • human locomotion principles with biomechanic simulation models
  • how to setup and control a dynamic system in real-time
  • how to control a simple system to demonstrate human locomotion principles in hardware.

In the first part, students will learn how to analyze human locomotion (e.g. hopping, walking, running) with ground reaction forces and joint kinematics based on inverse dynamics approach. Then forward dynamic modelling approach (spring loaded inverted pendulum) will be introduced to describe and explain human locomotion.

In the second part, students will learn how to control a single actuator (e.g. electric motor, pneumatic artificial muscle) system in real-time. This includes dynamic system modelling, basic control theory, sensor and actuator selection and real-time control implementation (e.g. Arduino, Matlab Real-Time).

In the last part, with the knowledge acquired from human biomechanics and dynamic system control, students will setup a simple system and demonstrate the human movement principles. For example, students could build an elbow soft exoskeleton to assist human arm movement.

xPC EtherCat system
xPC EtherCat system

Previous Projects

APRIL 2017

Laura Bielenberg, Jan Scheibe:

Control of the elbow joint angle in the vertical plane using Functional Electrostimulation (FES)

In this project, we investigated the accuracy of Functional electrostimulation (FES) for the control of elbow joint movements in the vertical plane. A crucial difficulty regarding this task, is the nonlinearity of the muscles torque gain relationship, making it complicated to model the muscles behaviour as an actuator. Prior research in the field of joint angle control with functional electrical stimulation (FES) showed, that the elbow angle can be regulated in using FES by implementing a cascaded feedback control. Working in the vertical plane – and therefore having to take gravitational force into account – we implemented a pilot control, which would calculate the stimulation – needed at that specific arm angle – from the FES-unit to generate the desired movement as accurate as possible. Subsequently, as well the control loop with, as also that without pilot control was tested. In the end we could show, that control of an elbow joint angle by FES is possible in a range of 15-90 degrees with a standard deviation of the angle error lower than 5 degree. However, against what the simulation suggested, the pilot control did not increase the systems stability; the standard angle deviation remained quite the same as with PI control only.

Daniel András, Alexander Breunig:

Design of a technical demonstration model for a finger prosthesis

In the following work the possibilities for a finger prosthesis were explored since the loss of a finger or parts of a finger leads to a major impairment. Most current finger prosthesis are passive and can’t be moved voluntarily. During this project, a robotic finger was designed. Right now, it can’t be used as a prosthesis but serves as a demonstrator regarding which sensors need to be implemented in a prosthesis and which controls are required. The demonstrator is equipped with a position control for the big movement and force sensors with the corresponding software control to enable precise gripping. It was shown, that even with basic equipment a precise movement can be achieved. The remaining problems are the miniaturization of the mechanical components and especially the power supply.

James Murphy, Andrej Scherf:

Design of a robotic soccer leg

This study describes the design and development of a biomechanical robot that was created as a final project for the module “Real time control of actuated systems”. The robotic extension, built onto an existing test rig, was used to extend the kicking range of a simple pendulum lower-limb design from 0.75m to 3.98m. The design and development of both the mechanical leg extension and the Arduino Simulink Real-time controller used to control actuation are described and the results of tests investigating the influence of spring stiffness and the pre-load force are discussed. The report also provides suggestions for further development of the lower limb extension and describes how it can be used to demonstrate the mechanics of serial elastic actuated joints.

David Dahlem, Victorio Lato:

Real Time Control of Operation Scissors with Force Feedback

In this project a real-time scissors platform has been successfully build. The scissors are motivated by real time operation scissors but are not designed for operational purposes. Moreover, the platform is kept more abstract with a focus on the control part. For manufacturing reasons the scissors are made of wood and are not designed to cut. The scissors can be opened and closed with a game controller. The system is controlled with an Arduino where a P controller for the position of the scissor angle is implemented. The force on the scissors is measured with a force sensor and given back to the user with a vibration feedback. The system follows given trajectories very fast and force feedback can already sense small forces.

Christian Alliger, Benjamin Northe, Felix Menzel:

Assistive biceps device and its improvements regarding muscle fatigue

Electromyography (EMG) signals are a promising way to couple humans to machines in real-time. This project focused on acquiring EMG signals from the biceps brachii, using these signals to control an assistive biceps device and comparing muscle fatigue with and without the assistance. This could be helpful for people who have muscle deficiencies but still have measureable EMG signals. A DC motor was used as an actuator to support the biceps. Steel Bowden cables and Velcro bands transmitted the forces to the body. The mean activation and the power spectral densities median frequency were used to evaluate fatigue. Real time data processing was a challenge because of the need of high filter orders due to the noisy signal resulting in unwanted high delays. The constructive part of this project was more a proof of concept than a user friendly one. The assistive device produces a notable support as long the EMG signals are not too noisy.