Program

Title: Intuitive robot programming from few demonstrations

Abstract: In contrast to other machine learning applications, robot learning typically relies on only few demonstrations or trials. Thus, the main challenge is to find structures and priors that can be used in a wide range of tasks. To reduce the amount of required data, a first opportunity to seize is that skills acquisition in robotics is a scaffolding process rather than a standard machine learning process. Indeed, we can exploit a number of interactive learning mechanisms to acquire/generate better data on-the-spot, by relying on social mechanisms to transfer skills more efficiently, including active learning, bilateral interactions, and kinesthetic teaching.

To encode skills in a compact and modular manner, movement primitives can also be used in robot learning as high-level "bricks" of motion that can be re-organized in series and in parallel. This notion can be extended to behavior primitives, forming a richer set of time-dependent and time-independent behaviors. I will show that this combination of primitives can be formalized as a product of experts (PoE), a machine learning technique modeling a probability distribution by combining the output of several simpler distributions. This fusion approach allows robots to counteract perturbations that have an impact on the fulfillment of the task, while ignoring other perturbations. This approach creates bridges with research in biomechanics and motor control, including minimal intervention principles, uncontrolled manifolds or optimal feedback control.

To facilitate the acquisition of manipulation skills, task-parameterized models can also be exploited to take into account that robot motions typically relate to objects, tools or landmarks in the robot's workspace. The approach consists of encoding a motion in multiple coordinate systems (e.g., from the perspectives of different objects), in the form of trajectory distributions. In a new situation (e.g., for new object locations), the reproduction challenge corresponds to a fusion problem, where the variations in the different coordinate systems are exploited to generate a movement reference tracked with variable gains, providing the robot with a variable impedance behavior that automatically adapts to the precision required in the different phases of the task. For example, in a pick-and-place task, the robot will be stiff if the object needs to be reached/dropped in a precise way, and will remain compliant in the other parts of the task.

The last part of my presentation will discuss how skills transfer can exploit stiffness and manipulability ellipsoids, in the form of geometric descriptors representing the skills to be transferred to the robot. As these ellipsoids lie on symmetric positive definite manifolds, the use of Riemannian geometry will be proposed as a way to learn and reproduce these descriptors in a probabilistic manner.

Video recording

Title: Admittance control and Control Barrier Functions for safe human-robot interaction

Abstract: Due to the increasing interest in pHRI, new safety standards have been published, in order to address human-robot collaboration in the industrial scenario. The risk in the mere application of the safety standards is to overconstrain the system with a final result of poor overall performance. In this talk, strategies for implementing safe pHRI applications by exploiting the Control Barrier Functions framework and the admittance control theory will be presented and discussed. The final objective is to find the best trade-off between safety and performance, by implementing advanced control algorithms.

Video recording

Title: Ergonomic control of human-robot co-manipulation

Abstract: The talk will present several robot control and learning methods for co-manipulation with humans, where the key element is an optimization of human ergonomics. The optimization and learning process incorporates biomechanical models and real-time measurements to track and improve various metrics, such as muscle fatigue, joint torques, and arm manipulability. In the first part, the focus is on the application to co-manipulation during physical human-robot collaboration in various practical tasks (e.g., collaborative sawing, polishing, valve turning, etc.). The second part will focus on the application in exoskeletons, where co-manipulation is done while human and robot limbs are physically coupled. Finally, the last part will examine the application in teleoperation, where co-manipulation pertains to the remote robot being commanded by a human operator. In particular, we will present an analysis of impedance-command interfaces in force-feedback tele-impedance.

Video recording

Title: Robots working with and around people

Abstract: Robots have begun to transition from assembly lines, where they are physically separated from humans, to environments where human–robot interaction is inevitable. With this shift, research in physical human–robot interaction (pHRI) has grown to allow robots to work with and around humans on complex tasks. Safe pHRI requires robots to both avoid harmful collisions and continue to work toward their main task whenever possible. Furthermore, robots must reliably sense their surroundings and parse pertinent information in real time to avoid potentially harmful collisions. However, as HRI scenarios become commonplace, resorting to pure collision avoidance is not sufficient – contact is inevitable, and, at times, desirable. In my talk, I will review my group's work on close-proximity HRI, i.e. allowing robots to avoid obstacles, to anticipate eventual collisions, and (as a longer-term objective) to purposely seek touch. I will present recent research on a framework capable of nearby-space perception, just-in-time control, and touch-informed motion planning, with the goal of creating "whole-body awareness" of the robot's surroundings. In all, this research will enable robot capabilities that were not previously possible, and will impact work domains other than manufacturing, including construction, logistics, and home care.

Video recording

Title: Interactive Learning for (Mobile) Manipulation

Abstract: Learning to solve complex manipulation tasks from visual observations is a dominant challenge for real-world robot learning. In this talk, I will first present our CEILing (Corrective and Evaluative Interactive Learning) framework that combines both corrective and evaluative feedback from a human to train a stochastic policy in an asynchronous manner, and employs a dedicated mechanism to trade off human corrections with the robot’s own experience. I will then present our deep reinforcement learning approach to learn feasible dynamic motions for a mobile robotic base while the end-effector follows a trajectory in task space generated by an arbitrary system to fulfill the task at hand. Finally, I will also introduce our on-going work on fairness-aware learning for safe human-robot interaction.

Video recording

Title: Towards industry 5.0 with collaborative robots

Abstract: Human-robot collaboration has great potential to face societal challenges (as ageing population, need for better and healthier work) and create new economic markets. In this talk we will focus how a smart design of the body, model based control and use of artificial intelligence, a new generation of collaborative robots (exoskeletons, cobots, soft grippers) can be developed for industry 5.0. Moreover by introducing self-healing materials, soft robots are produced that can heal damages, contributing to more sustainable robots. In these research projects is extensively collaborated with not only other technical fields as engineering, material sciences and computer sciences (AI), but also medical, human and social sciences. At the VUB we started the Brussels Human Robotics Research Center, BruBotics, which is a joint initiative of 8 research groups of the Vrije Universiteit Brussel (VUB) sharing a common vision: improve our quality of life through Human centered Robotics. Therefor we also introduce Homo Roboticus, to keep the human values central in a robotized world.

Video recording

Title: Safety in human-robot collaboration: state of the art and new perspectives

Abstract: The problem of guaranteeing safety of human workers has been a longstanding one in the robotics community for more than a decade. The availability of collaborative robotic solutions is further pushing towards the advancement in this field, as the number of factory installations is constantly growing. In this talk, a critical review of well-known possibilities available from industrial standards will be given, together with their possible limitations and challenges. In addition, novel paradigms developed at Politecnico di Milano will be presented to show their capability to outperform state of the art methods in terms of productivity.

Video recording