Publication: Stabilizer architecture for humanoid robots collaborating with humans
Loading...
Identifiers
Publication date
2012-02
Defense date
2012-06-20
Authors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Hoy en día, los avances en las tecnologías de información y comunicación permiten el uso de robots como compañeros en las actividades con los seres humanos. Mientras que la mayoría de las investigaciones existentes se dedica a la interacción entre humanos y robots, el marco de esta investigación está centrado en el uso de robots como agentes de colaboración. En particular, este estudio está dedicado a los robots humanoides que puedan ayudar a la gente en varias tareas en entornos de trabajo. Los robots humanoides son sin duda los m as adecuados para este tipo de situaciones: pueden usar las mismas herramientas que los seres humanos y son lo m as probablemente aceptados por ellos. Después de explicar las ventajas de las tareas de colaboración entre los humanos y los robots y las diferencias con respecto a los sistemas de interacción y de teleoperación, este trabajo se centra en el nivel de las tecnologías que es necesario para lograr ese objetivo. El problema más complicado en el control de humanoides es el balance de la estructura. Este estudio se centra en técnicas novedosas para la estimación de la actitud del robot, que se utilizarán para el control. El control del robot se basa en un modelo muy conocido y simplificado: el péndulo invertido. Este modelo permite tener un control en tiempo real sobre la estructura, mientras que esté sometida a fuerzas externas / disturbios. Trayectorias suaves para el control de humanoides se han propuesto y probado en plataformas reales: éstos permiten reducir los impactos del robot con su entorno. Finalmente, el estudio extiende estos resultados a una contribución para la arquitectura de colaboración humano-humanoide. Dos tipos de colaboraciones humano humanoide se analizan: la colaboración física, donde robots y humanos comparten el mismo espacio y tienen un contacto físico (o por medio de un objeto), y una colaboración a distancia, en la que el ser humano está relativamente lejos del robot y los dos agentes colaboran por medio de una interfaz. El paradigma básico de esta colaboración robótica es: lo que es difícil (o peligroso) para el ser humano se hace por medio del robot y lo que es difícil para el robot lo puede mejor hacer el humano. Es importante destacar que el contexto de los experimentos no se basa en una unica plataforma humanoide; por el contrario, tres plataformas han sido objeto de los experimentos: se han empleado los robots HOAP-3, HRP-2 y TEO. ----------------------------------------------------------------------------------------------------------------------------------------------------------
Nowadays, the advances in information and communication technologies permit the use of robots as companions in activities with humans. While most of the existing research is dedicated to the interaction between humans and robots, the framework of this research is the use of robots as collaborative agents. In particular, this study is dedicated to humanoid robots which should assist people in several tasks in working environments. Humanoid robots are certainly the most adequate for such situations: they can use the same tools as humans and are most likely accepted by them. After explaining the advantages of collaborative tasks among humans and robots and the differences with respect to interaction and teleoperation systems, this work focuses on the level of technologies which is necessary in order to achieve such a goal. The most complicated problem in humanoid control is the structure balance. This study focuses in novel techniques in the attitude estimation of the robot, to be used for the control. The control of the robot is based on a very well-known and simplified model: the double inverted pendulum. This model permits having a real-time control on the structure while submitted to external forces/disturbances. The control actions are strongly dependent on the three stability regions, which are determined by the position of the ZMP in the support polygon. Smooth trajectories for the humanoid control have been proposed and tested on real platforms: these permit reducing the impacts of the robot with its environment. Finally, the study extends these results to a contribution for human-humanoid collaboration architecture. Two types of human-humanoid collaborations are analyzed: a physical collaboration, where robot and human share the same space and have a physical contact (or by means of an object), and a remote collaboration, in which the human is relatively far away from the robot and the two agents collaborate using an interface. The basic paradigm for this robotic collaboration is: what is difficult (or dangerous) for the human is done by the robot and what is difficult for the robot is better done by the human. Importantly, the testing context is not based on a single humanoid platform; on the contrary, three platforms have been object of the experiments: the Hoap-3, HRP-2 and HRP2 robot have been employed.
Nowadays, the advances in information and communication technologies permit the use of robots as companions in activities with humans. While most of the existing research is dedicated to the interaction between humans and robots, the framework of this research is the use of robots as collaborative agents. In particular, this study is dedicated to humanoid robots which should assist people in several tasks in working environments. Humanoid robots are certainly the most adequate for such situations: they can use the same tools as humans and are most likely accepted by them. After explaining the advantages of collaborative tasks among humans and robots and the differences with respect to interaction and teleoperation systems, this work focuses on the level of technologies which is necessary in order to achieve such a goal. The most complicated problem in humanoid control is the structure balance. This study focuses in novel techniques in the attitude estimation of the robot, to be used for the control. The control of the robot is based on a very well-known and simplified model: the double inverted pendulum. This model permits having a real-time control on the structure while submitted to external forces/disturbances. The control actions are strongly dependent on the three stability regions, which are determined by the position of the ZMP in the support polygon. Smooth trajectories for the humanoid control have been proposed and tested on real platforms: these permit reducing the impacts of the robot with its environment. Finally, the study extends these results to a contribution for human-humanoid collaboration architecture. Two types of human-humanoid collaborations are analyzed: a physical collaboration, where robot and human share the same space and have a physical contact (or by means of an object), and a remote collaboration, in which the human is relatively far away from the robot and the two agents collaborate using an interface. The basic paradigm for this robotic collaboration is: what is difficult (or dangerous) for the human is done by the robot and what is difficult for the robot is better done by the human. Importantly, the testing context is not based on a single humanoid platform; on the contrary, three platforms have been object of the experiments: the Hoap-3, HRP-2 and HRP2 robot have been employed.
Description
Keywords
Human-Robot Interaction, Humanoid robots