Service Robots and Assistance Systems

Nr. Titel SWS Datum Raum Beginn
INF-61-81-L-7 Master-Projekt: Service Robots and Assistance Systems PT (halbtägig)

Mittwoch 14:30 - 15:30




Prof. Karsten Berns


M.Sc. Tobias Groll,, raum 48-352


Nach dem Anmelden finden sich hier Materialien zur Vorlesung und zu den Übungen.


Ziel des Projektes ist es ein Thema selbständig zu bearbeiten. Dies beeinhaltet die Planung und Implementierung von (Teil-) Systemen innerhalb unserer laufenden Projekte

Die Vorbesprechung für das Masterprojekt findet am Mittwoch 22.04.2015 um 14:30 Uhr in Raum 48-379 statt.

Im Sommersemester 2015 stehen folgende Themen zur Auswahl:

TitelKurze BeschreibungBetreuer
Optimization of the Formal Model of BBS for Verification The Robotics Research Lab conducts research in the field of behaviour-based robot control architectures (BBS) with a focus on the verification of such systems. At the RRLab, BBS are modelled as finite-state machines. Queries about properties of these networks are answered with the help of a model checker. A problem with the current techniques is the large state space which leads to long computing times and very high memory consumption. The task of this work is to optimize the currently used models of the different behaviours and to adapt the automatic modelling process to the new models. Lisa Kiekbusch
Extensions to the GUI for Analysis of BBS The Robotics Research Lab conducts research in the field of behaviour-based robot control architectures (BBS) with a focus on the analysis and verification of such systems. As several tools and techniques are used for the analysis, graphical support is very important to guide the user through the process. In the RRLab, the Finstruct tool visualizes networks of behaviours and provides some functionality for the verification. The task of this work is to add further analysis methods to the tool and to increase the overall handling. Lisa Kiekbusch
Simplified Soil Simulation for Simulated Truck Loading A simplified simulation is to be developed which simulates dropping of a specific amount of material above the truckbed and generates the resulting heightfield. Daniel Schmidt
Transferring the existing thor simulation environment including the excavator to V-Rep Model the excavator itself and model the environment in V-Rep and connect it to the existing control part. Also create new scenarios that are possible for testing. Daniel Schmidt
Interaction modules between Finroc and V-Rep We need special modules with typed interfaces to interact with elements in the V-Rep Simulation environment like heightfields, setting object poses, gathering sensor data, ... Daniel Schmidt
Hand gestures recognition using RGB-D Salah Saleh
Person IdentificationSalah Saleh
Simulation of a single Joint driven by linear Series Elastic Actuators in VRep Steffen Schütz
Building of a Linux distribution for an Altera Cyclone V SoC using PTXdist Steffen Schütz
Performance considerations of FPGA-based CORDIC implementations Steffen Schütz
Trajectory planning for LUGV Arm. Atabak Nezhadfard
single chamber control for CREA. Atabak Nezhadfard
Reinforcement learning for bipedal locomotion on rough terrain Qi Liu
Interface to interact with V-Rep from Finroc implemented as an V-Rep Plugin Tobias Groll
Local Path Planning for the 855D Gator* Patrick Fleischmann
Terrain Classification using an IMU Patrick Fleischmann
Tree Detection from Aerial Images for Robotic Mapping Patrick Fleischmann
Embedding/Implementation of the Reflexxes Motion Library in Finroc to control a backhoe arm Steffen Hemer
Tracking and Position Prediction Trajectory tracking is one basic task an autonomous vehicle should be capable of. A more sophisticated task is to follow a leader. Especially in nearby following scenarios, the detected leader position is often not sufficient. Moreover, if the leader islost, a search strategy has to be used to close up. Therefore, the effects of additional information from the leader like its orientation, steering angle and velocity shall be examined with respect to the tracking quality. A kinematic/dynamic model of the leading vehicle has to be created. Based on the latest information, the model shall be used to generate a trajectory lookahead. Thorsten Ropertz
Behavior-Based Risk Assessment In the Robotics Research Lab, the biologically motivated control approach IB2C is developed. One basic property of biological systems is their self-preservation instinct. Concerning robots this property is also very important in order to prevent harm from the robot and its environment. By now, basic collision avoidance mechanisms have been implemented based on IB2C, but a holistic risk assessment based on the current system state is missing. In context of this work, different risk types shall be identified and categorized. Additionally, features that allow for sensing and predicting the risk have to be determined and a behavior based network has to be created. Thorsten Ropertz
Dynamics Estimation In the order to control a complex and highly dynamic robot, an accurate dynamic model of the system is required. It serves as a required foundation for developing state of the art control algorithms as well as for verifying the control system in combination with the real hardware. Unfortunately, the dynamic model is usually not provided by the manufacturer such that the details have to be determined by testing and observing the vehicle behavior. In context of the project, the important properties shall be identified and a corresponding test suit shall be developed that allows for an easy estimation. Thorsten Ropertz
Design of Carbon Fiber stiffness reinforcement for chassis of CREA. Atabak Nezhadfard

*: Team work is possible for this topic





  • schriftliche Ausarbeitung
  • Präsentation