Walking paths during collaborative carriages do not follow the simple rules observed in the locomotion of single walking subjects.

Some works have already studied human trajectories during spontaneous locomotion. However, this topic has not been thoroughly studied in the context of human-human interactions, especially during collaborative carriage tasks. Thus, this manuscript aims to provide a broad analysis of the kinematics of two subjects carrying a table. In the present study, 20 pairs of subjects moved a table to 9 different goal positions distant of 2.7-5.4 m. This was performed with only one or both subjects knowing the target location. The analysis of the collected data demonstrated that there is no shared strategy implemented by all the pairs to move the table around. We observed a great variability in the pairs' behaviours. Even the same pair can implement various strategies to move a table to the same goal position. Moreover, a model of the trajectories adopted by collaborating pairs was proposed and optimized with an inverse optimal control scheme. Even if it produced consistent results, due to the great variability which origins were not elucidated, it was not possible to accurately simulate the average trajectories nor the individual ones. Thus, the approach that was shown to be efficient to simulate single walking subjects failed to model the behaviour of collaborating pairs.

Inverse optimal control to model human trajectories during locomotion.

Cobotic applications require a good knowledge of human behaviour in order to be cleverly, securely and fluidly performed. For example, to make a human and a humanoid robot perform a co-navigation or a co-manipulation task, a model of human walking trajectories is essential to make the robot follow or even anticipate the human movements. This paper aims to study the Center of Mass (CoM) path during locomotion and generate human-like trajectories with an optimal control scheme. It also proposes a metric which allows to assess this model compared to the human behaviour. CoM trajectories during locomotion of 10 healthy subjects were recorded and analysed as part of this study. Inverse optimal control was used to find the optimal cost function which best fits the model to the measurements. Then, the measurements and the generated data were compared in order to assess the performance of the presented model. Even if the experiments show a great variability in human behaviours, the model presented in this study gives an accurate approximation of the average human walking trajectories. Furthermore, this model gives an approximation of human locomotion good enough to improve cobotic tasks allowing a humanoid robot to anticipate the human behaviour.

How do walkers behave when crossing the way of a mobile robot that replicates human interaction rules?

Previous studies showed the existence of implicit interaction rules shared by human walkers when crossing each other. Especially, each walker contributes to the collision avoidance task and the crossing order, as set at the beginning, is preserved along the interaction. This order determines the adaptation strategy: the first arrived increases his/her advance by slightly accelerating and changing his/her heading, whereas the second one slows down and moves in the opposite direction. In this study, we analyzed the behavior of human walkers crossing the trajectory of a mobile robot that was programmed to reproduce this human avoidance strategy. In contrast with a previous study, which showed that humans mostly prefer to give the way to a non-reactive robot, we observed similar behaviors between human-human avoidance and human-robot avoidance when the robot replicates the human interaction rules. We discuss this result in relation with the importance of controlling robots in a human-like way in order to ease their cohabitation with humans.

Benchmarking the HRP-2 Humanoid Robot During Locomotion.

In this paper we report results on benchmarking a HRP-2 humanoid robot. The humanoid robots of this serie are known to be very robust. They have been successfully used by several research groups for the design of new motion generation algorithms. As such it is a reference in the category of electrically driven humanoid robot. As new humanoid robots are continuously built it is interesting to compare the performances of these new prototypes to those of HRP-2. This benchmarking study was realized through a campaign of measurements in an advanced equipped testing laboratory that provides a well adapted controlled environment. We have investigated the effect of temperatures variation on the robot walking capabilities. In order to benchmark various environmental conditions and algorithms we computed a set of performance indicators for bipedal locomotion. The scope of the algorithms for motion generation evaluated here ranges from analytical solution to numerical optimization approach, enabling real-time walking or multi-contacts motions.

How do walkers avoid a mobile robot crossing their way?

Robots and Humans have to share the same environment more and more often. In the aim of steering robots in a safe and convenient manner among humans it is required to understand how humans interact with them. This work focuses on collision avoidance between a human and a robot during locomotion. Having in mind previous results on human obstacle avoidance, as well as the description of the main principles which guide collision avoidance strategies, we observe how humans adapt a goal-directed locomotion task when they have to interfere with a mobile robot. Our results show differences in the strategy set by humans to avoid a robot in comparison with avoiding another human. Humans prefer to give the way to the robot even when they are likely to pass first at the beginning of the interaction.

Reverse control for humanoid robot task recognition.

Efficient methods to perform motion recognition have been developed using statistical tools. Those methods rely on primitive learning in a suitable space, for example, the latent space of the joint angle and/or adequate task spaces. Learned primitives are often sequential: A motion is segmented according to the time axis. When working with a humanoid robot, a motion can be decomposed into parallel subtasks. For example, in a waiter scenario, the robot has to keep some plates horizontal with one of its arms while placing a plate on the table with its free hand. Recognition can thus not be limited to one task per consecutive segment of time. The method presented in this paper takes advantage of the knowledge of what tasks the robot is able to do and how the motion is generated from this set of known controllers, to perform a reverse engineering of an observed motion. This analysis is intended to recognize parallel tasks that have been used to generate a motion. The method relies on the task-function formalism and the projection operation into the null space of a task to decouple the controllers. The approach is successfully applied on a real robot to disambiguate motion in different scenarios where two motions look similar but have different purposes.

MonoSLAM: real-time single camera SLAM.

We present a real-time algorithm which can recover the 3D trajectory of a monocular camera, moving rapidly through a previously unknown scene. Our system, which we dub MonoSLAM, is the first successful application of the SLAM methodology from mobile robotics to the "pure vision" domain of a single uncontrolled camera, achieving real time but drift-free performance inaccessible to Structure from Motion approaches. The core of the approach is the online creation of a sparse but persistent map of natural landmarks within a probabilistic framework. Our key novel contributions include an active approach to mapping and measurement, the use of a general motion model for smooth camera movement, and solutions for monocular feature initialization and feature orientation estimation. Together, these add up to an extremely efficient and robust algorithm which runs at 30 Hz with standard PC and camera hardware. This work extends the range of robotic systems in which SLAM can be usefully applied, but also opens up new areas. We present applications of MonoSLAM to real-time 3D localization and mapping for a high-performance full-size humanoid robot and live augmented reality with a hand-held camera.