A Modular Approach to Learning Manipulation Strategies from Human Demonstration

Object manipulation is a challenging task for robotics, as the physics involved in object interaction is com- plex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human perform- ing a task that requires an adaptive control strategy in differ- ent conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each mod- ule represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated er- ror in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.

A modular approach to learning manipulation strategies from human demonstration

Aude Billard, Ravin Luis De Souza, Miao Li

Object manipulation is a challenging task for robotics, as the physics involved in object interaction is complex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human performing a task that requires an adaptive control strategy in different conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each module represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated error in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.

Springer Verlag2016

Control and Learning of Compliant Manipulation Skills

Klas Jonas Alfred Kronander

Humans demonstrate an impressive capability to manipulate fragile objects without damaging them, graciously controlling the force and position of hands or tools. Traditionally, robotics has favored position control over force control to produce fast, accurate and repeatable motion. For extending the applicability of robotic manipulators outside the strictly controlled environments of industrial work cells, position control is inadequate. Tasks that involve contact with objects whose positions are not known with perfect certainty require a controller that regulates the relationship between positional deviations and forces on the robot. This problem is formalized in the impedance control framework, which focuses the robot control problem on the interaction between the robot and its environment. By adjusting the impedance parameters, the behavior of the robot can be adapted to the need of the task. However, it is often difficult to specify formally how the impedance should vary for best performance. Furthermore, fast it can be shown that careless variation of the impedance can lead to unstable regulation or tracking even in free motion. In the first part of the thesis, the problem of how to define a varying impedance for a task is addressed. A haptic human-robot interface that allows a human supervisor to teach impedance variations by physically interacting with the robot during task execution is introduced. It is shown that the interface can be used to enhance the performance in several manipulation tasks. Then, the problem of stable control with varying impedance is addressed. Along with a theoretical discussion on this topic, a sufficient condition for stable varying stiffness and damping is provided. In the second part of the thesis, we explore more complex manipulation scenarios via online generation of the robot trajectory. This is done along two axes 1) learning how to react to contact forces in insertion tasks which are crucial for assembly operations and 2) autonomous Dynamical Systems (DS) for motion representation with the capability to encode a family of trajectories rather than a fixed, time-dependent reference. A novel framework for task representation using DS is introduced, termed Locally Modulated Dynamical Systems (LMDS). LMDS differs from existing DS estimation algorithms in that it supports non-parametric and incremental learning all the while guaranteeing that the resulting DS is globally stable at an attractor point. To combine the advantages of DS motion generation with impedance control, a novel controller for tasks described by first order DS is proposed. The controller is passive, and has the properties of an impedance controller with the added flexibility of a DS motion representation instead of a time-indexed trajectory.

EPFL2015

Enactive robot vision

Mototaka Suzuki

The complexity of today's autonomous robots poses a major challenge for Artificial Intelligence. These robots are equipped with sophisticated sensors and mechanical abilities that allow them to enter our homes and interact with humans. For example, today's robots are almost all equipped with vision and several of them can move over rough terrain with wheels or legs. The methods developed so far in Artificial Intelligence, however, are not yet ready to cope with the complexity of the information gathered through the robot sensors and the need for rapid action in partially unknown and dynamic environments. In this thesis, I will argue that the apparent complexity of the environment and of the robot brain can be significantly simplified if perception, behavior, and learning are allowed to co-develop on the same time scale. In doing so, robots become sensitive to, and actively exploit, characteristics of the environment that they can tackle within their own computational and physical constraints. This line of work is grounded on philosophical and psychological research showing that perception is an active process mediated by behavior. However, computational models of active vision are very rare and often rely on architectures that are preprogrammed to detect certain characteristics of the environment. Previous work have shown that complex visual tasks, such as position and size invariant shape recognition as well as navigation, can be tackled with remarkably simple neural architectures generated by a coevolutionary process of active vision and feature selection. Behavioral machines equipped with primitive vision systems and direct pathways between visual and motor neurons were evolved while freely interacting with their environments. I proceed further on this line of investigation and describe the application of this methodology in three situations, namely car driving with an omnidirectional camera, goal-oriented navigation of a humanoid robot, and cooperative tasks by two agents. I will show that these systems develop sensitivity to a number of oriented, retinotopic, visual features – oriented edges, corners, height – and a behavioral repertoire to locate, bring, and keep these features in sensitive regions of the vision system that allow them to accomplish their goals. In a second set of experiments, I will show that active vision can be exploited by the robot to perform anticipatory exploration of the environment in a task that requires landmark-based navigation. Evolved robots exploit an internal expectation system that makes use of active exploration to check for expected events in the environment. I will describe a third set of experiments where, in addition to an evolutionary process, the visual system of the robot can develop receptive fields by means of unsupervised Hebbian learning and show that these receptive fields are significantly affected by the behavior of the system and differ from those predicted by most computational models of visual cortex. Finally, I will show that these robots replicate the performance deficiencies observed in experiments of motor deprivation with kitten when they are exposed to the same type of motor deprivations. Furthermore, the analyses of our robot brains suggest an explanation for the deficiencies observed in kitten that have not yet been fully understood.