Learning from Demonstration (LfD) aims to learn versatile skills from human demonstrations. The field has been gaining popularity since it facilitates transferring knowledge to robots without requiring much expert knowledge. During task executions, the robot motion is usually influenced by constraints imposed by environments. In light of this, task-parameterized (TP) learning encodes relevant contextual information in reference frames, enabling better skill generalization to new situations. However, most TP learning algorithms require multiple demonstrations in various environment conditions to ensure sufficient statistics for a meaningful model. It is not a trivial task for robot users to create different situations and perform demonstrations under all of them. Therefore, this paper presents a novel concept to learn motion policy from few demonstrations through explicitly solving reference frame weights along the task trajectory. Experimental results in both simulation and real robotic environments validate our approach.

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 to find a suitable representation with low dimensionality and reliable accuracy. The aim of this thesis to develop an algorithm to represent the state of the deformable objects like cloth in low-dimensional vectors, together with a framework based on visual servoing to flatten cloth-like objects. We present a novel pipeline for cloth flattening, which determines a stretching direction (in 2D vector) and an operation point for the robot to removes the wrinkles. The performance of the perception algorithm is validated in simulation and real-world experiment. The whole framework is evaluated in the real-world experiment, which is compared with a human operator. The results show that our framework efficiently determines the direction of wrinkles on the cloth in the simulation as well as the real robot experiment. Besides, the proposed framework has a good performance close to that of a human operator in terms of cloth flattening tasks.

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 visual perception and human demonstrations. We present a novel feature, Wrinkle cOntRaction Direction (WORD), which extracts a stretching direction to flatten clothes from image and depth data. Furthermore, we integrate WORD and demonstrations into Gaussian Processes to learn a cloth placing policy. Simulation and robot experiment results are used to validate the performance of WORD and the proposed learning framework in this study. The results show that WORD efficiently captures wrinkles on the contact part of the cloth in the simulation as well as the real robot experiment. Besides, the proposed learning framework performs successful results in cloth placing and flattening.A video of the experiments and execution of the tasks is available at https://youtu.be/iV2mAPqL7mA.