Autonomous Active Object Learning Through Robot Manipulation

服务机器人有可能成为我们日常生活中无处不在的帮手。他们被设想执行各种任务，从打扫我们的公寓到修剪我们的草坪或喂养我们的宠物。为了成功地解决这类任务，机器人必须能够正确地推理和操作对象，这需要丰富的关于相应对象的知识。为了获得这样的知识，机器人不仅需要分析对象的几何和外观，还需要检索对象的类别和推断对象的特征。这包括物理属性，如材料和重量，或部件及其功能，以及其他形式的信息，如物体品牌或价格。在这个项目中，我们的目标是通过利用物理物体相互作用的可能性，并通过分析万维网上可用的信息，在最少的人工监督下了解所有物体信息。此外，我们的目标是将操作技能从一般模型迁移到新对象的特定形状和特征以及这些对象的自主改进。在我们的研究中，我们将使用最先进的配备机械手的机器人。S。我们为机器人提供关于物体的先验知识，并让机器人访问网络上不断更新的、具有很大多样性的信息。机器人通过与环境的交互来测试所估计的假设来主动学习，例如通过获取对象数据、推动或拾取对象。它从演示中学习操作对象的新技能，将其推广到功能相似的对象，并从这些对象的经验中改进。ALROMA将推动主动和机器人学习、对象操作和对象发现方面的最新水平，并将为下一代服务机器人做出贡献。

Service robots have the potential to be ubiquitous helpers of our everyday life. They are envisioned to perform a variety of tasks ranging from cleaning our apartments to mowing our lawn or feeding our pets. For solving such tasks successfully, robots must be able to properly reason about and manipulate objects, which requires rich knowledge about the corresponding objects. To obtain such knowledge, robots need not only means for analyzing the geometry and appearance of objects but also for retrieving their categories and inferring their characteristics. This includes, for example, physical properties such as material and weight, or parts and their functions, as well as other forms of information such as object brand or price.In this project, we aim at learning all the object information with minimal human supervision by leveraging the possibility of physical object interaction and by analyzing the information available in the World Wide Web. Furthermore, we aim at the transfer of manipulation skills from a generic model to the specific shape and characteristics of novel objects and the autonomous improvement of these.For our research, we will use state-of-the-art robots equipped with manipulatorWe propose a novel active learning perspective. s. We provide the robot with prior knowledge about objects and give the robot access to the information available on the web that is continually updated and with large diversity. The robot actively learns by testing the estimated hypotheses through interaction with the environment, e.g., by taking object data, pushing or picking an object. It learns novel skills for manipulating the objects from demonstrations, which it generalizes to functionally similar objects and improves from experience with these.ALROMA will advance the state of the art in active and robot learning, object manipulation and object discovery and will contribute to the next generation of service robots.