Benchmark for Human-to-Robot Handovers of Unseen Containers with Unknown Filling

The real-time estimation through vision of the physical properties of objects manipulated by humans is important to inform the control of robots for performing accurate and safe grasps of objects handed over by humans. However, estimating the 3D pose and dimensions of previously unseen objects using only RGB cameras is challenging due to illumination variations, reflective surfaces, transparencies, and occlusions caused both by the human and the robot. In this paper we present a benchmark for dynamic human-to-robot handovers that do not rely on a motion capture system, markers, or prior knowledge of specific objects. To facilitate comparisons, the benchmark focuses on cups with different levels of transparencies and with an unknown amount of an unknown filling. The performance scores assess the overall system as well as its components in order to help isolate modules of the pipeline that need improvements. In addition to the task description and the performance scores, we also present and distribute as open source a baseline implementation for the overall pipeline to enable comparisons and facilitate progress.

Learning How to Write with a Social Robot

Shruti Chandra

Learning handwriting is one of the main occupations for children in primary education. It is a complex skill that has been associated with children’s learning and development. Poor handwriting skills affect children negatively concerning their academic performance and motivation. In schools, children with handwriting problems need extra support and one-to-one interventions from teachers. Personalised one-to-one support requires additional resources that most schools generally cannot afford. Therefore, there is a need to explore different methods of learning and contemporary technologies to develop new educational tools to foster children’s learning process of handwriting. The research presented in this thesis addresses the question of - “How to create a social robot that can help children to acquire handwriting skills?". Thus, we looked into three different research domains: education; pedagogic sciences; and social human-robot interaction (HRI). On the one hand, we investigated the types of errors commonly present in children’s handwriting. On the other hand, we needed to explore different educational theories to find alternate learning methods that would maximise children’s learning gains. In addition, we investigated various possibilities for integrating a social robot into educational scenarios and measured its impact on the children’s learning process. This thesis presents five studies that follow an incremental evolution of the research. The first three studies were based on the interaction between two children, in the presence of either a human or a robotic facilitator, with the children performing a collaborative writing activity. We explored the ‘peer-learning’ and ‘peer-tutoring’ methods and compared the differences in terms of interaction behaviour and learning gains. Further, we analysed children’s interpersonal distance and feelings of responsibility towards their peers, both in the presence of a robot and human facilitator. Finally, we examined children’s assessment behaviour towards their peers and self-disclosure towards the robotic facilitator. The outcomes of these first studies provided enough data to generate a taxonomy of children handwriting errors to be used in the subsequent studies. They hinted that the ‘peer-tutoring’ method had a positive influence on children’s learning and corrective feedback to their peers in the presence of the social robot, providing positive support to the applicability of social robots as ‘facilitators’ in handwriting learning. The next step was to study the effects of the robot in the peer-tutoring scenario when assuming the role of a peer, directly interacting and performing a collaborative writing activity with the children. Thus, we developed an educational system with an autonomous social behaviour of a robot exhibiting various levels of learning competencies. The tutor-child provides corrective feedback on the writing of the learner-robot and in turn, enhances his writing skills. The robot’s writing was generated by an algorithm incorporating human-inspired movements and could reproduce a set of writing errors from the error taxonomy. We tested the system by conducting two longitudinal studies, each of them composed of four sessions. In the fourth study, we assigned the robot two contrasting competencies: ‘learning’ and ‘non-learning’. We measured the differences in children’s learning gains and changes in their perceptions towards the learner-robot. The fifth study followed a similar interaction scenario and research questions, but this time the robot performed three learning competencies: ‘continuous-learning’;’non-learning’ and ‘personalised-learning’. The findings of these studies show that the children improved more with the robot exhibiting learning, continuous-learning and personalised-learning competencies compared to the robot showing non-learning competency. The way children’s learning and perception of the robot evolved over time confirmed the need for multiple sessions for the last two studies. This thesis presents some evidence that social robots can be successfully used as tools for enhancing children’s handwriting skills in peer-learning scenarios.

EPFL2019

Using reinforcement learning for optimizing the reproduction of tasks in robot programming by demonstration

Florent Günter

As robots start pervading human environments, the need for new interfaces that would simplify human-robot interaction has become more pressing. Robot Programming by Demonstration (RbD) develops intuitive ways of programming robots, taking inspiration in strategies used by humans to transmit knowledge to apprentices. The user-friendliness of RbD is meant to allow lay users with no prior knowledge in computer science, electronics or mechanics to train robots to accomplish tasks the same way as they would with a co-worker. When a trainer teaches a task to a robot, he/she shows a particular way of fulfilling the task. For a robot to be able to learn from observing the trainer, it must be able to learn what the task entails (i.e. answer the so-called "What-to-imitate?" question), by inferring the user's intentions. But most importantly, the robot must be able to adapt its own controller to fit at best the demonstration (the so-called "How-to-imitate?" question) despite different setups and embodiments. The latter is the question that interested us in this thesis. It relates to the problem of optimizing the reproduction of the task under environmental constraints. The "How-to-imitate?" question is subdivided into two problems. The first problem, also known as the "correspondence problem", relates to resolving the discrepancy between the human demonstrator and robot's body that prevent the robot from doing an identical reproduction of the task. Even though we helped ourselves by considering solely humanoid platforms, that is platforms that have a joint configuration similar to that of the human, discrepancies in the number of degrees of freedom and range of motion remained. We resolved these by exploiting the redundant information conveyed through the demonstrations by collecting data through different frames of reference. By exploiting these redundancies in an algorithm comparable to the damped least square algorithm, we are able to reproduce a trajectory that minimizes the error between the desired trajectory and the reproduced trajectory across each frame of reference. The second problem consists in reproducing a trajectory in an unknown setup while respecting the task constraints learned during training. When the information learned from the demonstration no longer suffice to generalize the task constraints to a new set-up, the robot must re-learn the task; this time through trial-and-error. Here we considered the combination of trial-and-error learning to complement RbD. By adding a trial-and-error module to the original Imitation Learning algorithm, the robot can find a solution that is more adapted to the context and to its embodiment than the solution found using RbD. Specifically, we compared Reinforcement Learning (RL) – to other classical optimization techniques. We show that the system is advantageous in that: a) learning is more robust to unexpected events that have not been encountered during the demonstrations and b) the robot is able to optimize its own model of the task according to its own embodiment.

EPFL2009

Learning, Generating and Adapting Wave Gestures for Expressive Human-Robot Interaction

Sylvain Calinon

This study proposes a novel imitation learning approach for the stochastic generation of human-like rhythmic wave gestures and their modulation for effective non-verbal communication through a probabilistic formulation using joint angle data from human demonstrations. This is achieved by learning and modulating the overall expression characteristics of the gesture (e.g., arm posture, waving frequency and amplitude) in the frequency domain. The method was evaluated on simulated robot experiments involving a robot with a manipulator of 6 degrees of freedom. The results show that the method provides efficient encoding and modulation of rhythmic movements and ensures variability in their execution.