Intuitive programming of self-adaptive robots (ABOT)

A typical task requiring the use of external sensors for successful execution

Object locations are saved as probability distributions to optimise the search for the next execution

Abstract

The aim of this project is to define an intuitive way for the development of complex sensor-based robot programs. Nowadays this is a very time-consuming task and requires detailed knowledge about the type of application and the format of the sensor data. Because of this the use of sensors in modern robot-programs is still not very widespread.

The main idea in this project is to use the robots sensors to gain information about the environment during multiple executions of the task at hand. This knowledge will not be specified by the programmer but is gained autonomously by the robot itself when executing the task. The gained information is then used to adapt the programm automatically to the given task resulting in a faster and more robust execution the next time.

Our aim is to encapsulate this adaption-process in a simple, easy-to-use programming environment allowing programmers to generate complex, sensor-based robot programs without having to deal with the task of defining and processing the sensory information.

Applications

Sensor-supported robot programmes for industrial applications
Manipulation of flexible and alternating objects
Automatic adaptation to dynamic work environments

Contact

Prof. Dr. Dominik HENRICH

Pictures

Pictures for press usage

Press releases

Spezialisten überflüssig - Die Sensorik Integration — Wie lassen sich Roboter-Applikationen für Handhabungsaufgaben auch durch Nicht-Experten einfach programmieren? Das Projekt "A-Bot" zeigt neue Ansätze auf.
(Computer & AUTOMATION - Fachmagazin der Fertigungs- und Prozesstechnik, 07. Oktober 2009)