Controlling the estimation bias in deep reinforcement learning problems with sparse rewards: Towards robust robotic object manipulation learning

Many recent robot learning problems, real and simulated, were addressed using deep reinforcement learning. The developed policies can deal with high-dimensional, continuous state and action spaces, and can also incorporate machine-generated or human demonstration data. A great number of them depend on state-action value estimates, especially the...

Wrinkle direction detection and its application on robotic cloth wrinkle removal

Deformable Object Manipulation (DOM) is an important field of research as it contributes to practical tasks such as cloth handling, cable routing, surgical operation etc. The sensing in DOM is now considered as one of the major challenges in robotics due to the complex dynamics and high degree of freedom of deformable objects. One challenge is...

Cooperative Control of Autonomous Multi-Vessel Systems for Floating Object Manipulation

This thesis provides a set of cooperative control schemes for autonomous multi-vessel systems to manipulate a floating object through physical interconnections in onshore (inland waterways and ports) and offshore areas. Thanks to the maturity and popularity of the advancing technologies in information, communication, sensors, automatic control,...

Submillimeter lateral displacement enables friction sensing and awareness of surface slipperiness

Human tactile perception and motor control rely on the frictional estimates that stem from the deformation of the skin and slip events. However, it is not clear how exactly these mechanical events relate to the perception of friction. This study aims to quantify how minor lateral displacement and speed enables subjects to feel frictional...

Wrinkle contraction direction: a useful feature for learning robotic fabric manipulation from demonstration

Deformable objects manipulation (DOM) is largely considered an open problem in robotics. The complexity stems from the high degrees of freedom and nonlinear nature of the object configurations. In this thesis, we consider placing and flattening tasks for cloth-like objects. We propose a practical framework to place a cloth on a surface based on...

Single Shot Learning of the Stiffness Distribution of a Soft Object

With robotics rapidly expanding towards new user-oriented applications, such as agriculture, households, and classrooms, new tasks and new requirements arise. One of the ongoing and unsolved problems that comes with the unpredictable new environments robots find themselves in, is soft object manipulation. Without assuming the handled object to...