Volume 3, Issue 2

We present a database and a software tool, VisGraB, for benchmarking of methods for vision-based grasping of unknown objects with no prior object knowledge. The benchmark is a combined real-world and simulated experimental setup. Stereo images of real scenes containing several objects in different configurations are included in the database. The user needs to provide a method for grasp generation based on the real visual input. The grasps are then planned, executed, and evaluated by the provided grasp simulator where several grasp-quality measures are used for evaluation. This setup has the advantage that a large number of grasps can be executed and evaluated while dealing with dynamics and the noise and uncertainty present in the real world images. VisGraB enables a fair comparison among different grasping methods. The user furthermore does not need to deal with robot hardware, focusing on the vision methods instead. As a baseline, benchmark results of our grasp strategy are included.

This paper examines the impact of delays on human performance and human strategies when remotely navigating autonomous vehicles, and develops a robust human inspired delay compensation. Vehicles chosen for the study are ground autonomous vehicles which are allowed to stop, providing an instrumental feature that enables it to capture some important human behavior. The effects of delay on human behavior when remotely navigating autonomous vehicles have been captured by a nonlinear model predictive (also known as receding horizon) controller. This study provides some insights into designing human in-the-loop systems for remote navigation of autonomous vehicles when the delays are not negligible. We offer a human inspired strategy for dealing with delay in a fully autonomous receding horizon controller which we show to be safe and convergent for bounded delays.

In the last years of research in cognitive control, neuroscience and humanoid robotics have converged to different frameworks which aim, on one side, at modeling and analyzing human motion, and, on the other side, at enhancing motor abilities of humanoids. In this paper we try to cover the gap between the two areas, giving an overview of the literature in the two fields which concerns the production of movements. First, we survey computational motor control models based on optimality principles; then, we review available implementations and techniques to transfer these principles to humanoid robots, with a focus on the limitations and possible improvements of the current implementations. Moreover, we propose Stochastic Optimal Control as a framework to take into account delays and noise, thus catching the unpredictability aspects typical of both humans and humanoids systems. Optimal Control in general can also easily be integrated with Machine Learning frameworks, thus resulting in a computational implementation of human motor learning. This survey is mainly addressed to roboticists attempting to implement human-inspired controllers on robots, but can also be of interest for researchers in other fields, such as computational motor control.

It is a major challenge in HRI to design a social robot that is able to selectively direct a target human’s attention towards an intended direction. For this purpose, the robot may first turn its gaze toward him/her in order to establish eye contact. However, such a turning action of the robot may not in itself be sufficient to make eye contact with the target person in all situations, especially when the robot and the person are not facing each other or the human is intensely engaged in a task. In this paper, we propose a conceptual model of attention control with five phases: attention attraction, eye contact, attention avoidance, gaze back, and attention shift. Evaluation experiment by using a robotic head reveals the effectiveness of the proposed model in different viewing situations.

The capability to use voice commands to control robots offer an intriguing possibility to increase the efficiency with which robotic operators may give commands. This study consists of two experiments that investigate how robot operators prefer to speak to robots in a search-and-find task and to evaluate which mode of speaking generates the greatest performance. Experiment 1 revealed that operators used selective exocentric references when available to direct a confederate acting as a robot. Experiment 2 revealed that with the same exocentric references operators showed improved performance while directing an actual robot as compared with egocentric-only commands. Experiment 2 also revealed that performing a dual task was less detrimental to performance when using exocentric commands as compared to egocentric commands. Suggestions for improvements to a robot control system that follow from these results include developing recognition of structural properties of an indoor environment and improving map incorporation.