Introducing Q-Rock : Towards the Automated Self-Exploration and Qualification of Robot Behaviors
Thomas M Roehr, Daniel Harnack, Oscar Lima, Hendrik Wöhrle, Frank Kirchner
In Poster at the ICRA 2019 Workshop on Robot Design and Customization: Opportunities at the Intersection of Computation and Digital Fabrication, (ICRA-2019), 20.5.-24.5.2019, Montreal, Quebec, o.A., pages 1-2, May/2019.
Abstract
:
The increasing complexity of robotic systems de- mands novel approaches for their design and special support for their development and programming. To tackle this issue, the project Q-ROCK presents a novel, life-cycle inspired development process for the automated self-exploration and qualification of robotic capabilities and behaviors. The development process in Q- ROCK consists of three interdependent steps: (1) the automated exploration of capabilities that given robotic hardware can perform, (2) the classification and semantic annotation of these capabilities to generate more complex behaviors, and (3) the mapping of existing application requirements to a robot system that is capable of fulfilling these requirements due to its set of capabilities and behaviors. Each of the steps builds upon the results of previous steps. Furthermore, Q-ROCK 's main intention lies in the introduction of a unified development cycle comprising the selection, development and improvement of robots, which can be used by robot experts and novice end- users alike. By traversing the development cycle multiple times, a comprehensive, continuously growing knowledge-base including but not limited to hardware and software components, their capabilities and inter-dependencies is evolved.
Keywords
:
automated automated robot design,automated robot synthesis,behavior exploration,transfer learning
Files:
20190501_roehr_introducing_qrock.pdf