Human-Robot Co-Working is an active research domain in the field of robotics and control theory. Most of pioneering contributions addressing Co-Working interaction control problems has focused on PID based control. On the other hand, there are, in modern control theory, several tools allowing to design and tune arbitrarily complex servo-controllers. Non-Smooth optimization is one of recently developed tools that addresses the problem of tuning structured controllers according to H∞, H2, and pole placement specifications. In this paper, the problem of force amplification is transcribed into high level H∞specifications and is solved using Non-Smooth optimization. The resulting controller is tested on an industrial robot and shows similar performances as the state of the art while respecting some extra robustness certificates. Therefore, thanks to the abstraction level of the proposed technique, it becomes possible to tune interaction controllers with high number of tunable parameters according to complex performance/robustness specifications.

In co-manipulation, humans and robots solve manipulation tasks together. Virtual guides are important tools for co-manipulation, as they constrain the movement of the robot to avoid undesirable effects, such as collisions with the environment. Defining virtual guides is often a laborious task requiring expert knowledge. This restricts the usefulness of virtual guides in environments where new tasks may need to be solved, or where multiple tasks need to be solved sequentially, but in an unknown order. To this end, we propose a framework for multiple probabilistic virtual guides, and demonstrate a concrete implementation of such guides using kinesthetic teaching and Gaussian mixture models. Our approach enables non-expert users to design virtual guides through demonstration. Also, they may demonstrate novel guides, even if already known guides are active. Finally, users are able to intuitively select the appropriate guide from a set of guides through physical interaction with the robot. We evaluate our approach in a pick-and-place task, where users are to place objects at one of several positions in a cupboard.

It is well-known that iff the robot with its end-effector force control is passive, the closed loop system consisting of the robot and an arbitrary passive environment is asymptotically stable. Passive robot control, however, limits the achievable robot impedance reduction. Recently, investigated what performance can be achieved with nonpassive control when uncertainty bounds for the environment are known. The question of stability margins neglected in is treated in our companion paper. In this paper we report on experiments on a full scale industrial robot in order to identify six transfer functions from the velocity reference of the inner velocity loop to the end effector force output for six different environments, from solid wall to air. It is shown how the experimental data also makes it possible to compute the impedance of the controlled robot. Two controllers are designed: one marginally stable but with nonpassive and low-gain controlled robot impedance at low frequencies, and another controller respecting the stability margin of closed loop sensitivity gain less than 6 dB, but with a passive controlled robot impedance that has higher gain for low frequencies than that of the first controller. The designs are compared experimentally by having the robot interact with environments of different impedances, and with a human operator leading the robot The paper is concluded by a short discussion of the possible need of controller adaption when a robot operates in varying impedance environments, and interacts with a human operator.

In the field of comanipulation (i.e. a man and a robot sharing the same task), force amplification is an interesting function that can be achieved by using two force sensors. This technique is known in the literature but little attention has been paid so far to stability/passivity properties. We will explain how to deal with passivity based stability criteria, and point out performance limitations of such a control in case of noncollocated and bandwidth limited force sensors. Theoretical, as well as experimental results will be presented.

We focus on comanipulation, i.e. manipulation of an object simultaneously held by a robot and a human operator. In this domain, a major difficulty is raised by significant variations of human dynamics, which depend not only on the arm posture, but also on the muscular activity (muscular co-contraction) and more generally on the type of task being performed: fine positioning, gross and rapid movements, repeated movements, etc. An ideal comanipulation system should be able of adapting its behavior to the operator's functional intention, resulting in an intuitive assisting device. Toward this goal, we present in this paper first results of our research aimed at developing an instrumented handle mounted on a robot end-effector and held by an operator, that can be used for estimating the grasping force and for adapting the robot controller accordingly. We show first experimental evidences that changes in the grasping force drastically affect the robot controller performances. We thus propose a handle design and a gain scheduling strategy that result in a robot behavior adequate for any kind of grasps. This solution is successfully experimented with a 1 degree of freedom robot under largely variable comanipulation conditions, exhibiting a stable and efficiently adaptive behavior.