社交机器人：2012年国际会议论文集

需积分: 10 88 浏览量更新于2024-07-19 收藏 48.18MB PDF 举报

"Social Robotics"是一本专注于社会机器人学的书籍，收录了第四届国际社会机器人会议（ICSR2012）的论文集，该会议于2012年10月29日至31日在成都举行。这本书由Shuzhi Sam Ge、Oussama Khatib和John-John Cabibihan等人编辑，属于"Lecture Notes in Artificial Intelligence"系列，旨在探讨机器人技术在社交交互中的应用和最新发展。社会机器人学是机器人科学的一个分支，它涉及设计、构建和研究能够与人类进行有效交互的智能机器。这个领域涵盖了多个学科，包括人工智能、机械工程、心理学、社会学和人类行为学。书中可能包含了以下关键知识点： 1. **人机交互(Human-Robot Interaction, HRI)**：讨论如何设计机器人以便更好地理解人类的意图、情感和非语言信号，同时让机器人能够以自然、友好的方式回应。 2. **机器人感知(Pervasive Sensing)**：机器人通过各种传感器获取环境信息，如视觉、听觉和触觉，以实现与人的无缝沟通。 3. **自然语言处理(Natural Language Processing, NLP)**：使机器人能理解和生成人类语言，进行有效的口头交流。 4. **情感计算(Affective Computing)**：研究如何让机器人识别和响应人类的情绪，以增强人际互动的真实感。 5. **社会规范和社会规则(Social Norms and Etiquette)**：机器人需要了解并遵守人类社会的规则，以便在各种场合中适应得体。 6. **伦理和法律问题(Ethical and Legal Issues)**：随着社会机器人在日常生活中的应用增加，相关的伦理和法律挑战也日益突出，例如隐私权、责任归属等。 7. **可接受性研究(Acceptance Studies)**：探讨人们对于与机器人互动的接受程度，以及如何提高机器人的社会接受度。 8. **机器人学习(Learning in Robots)**：包括模仿学习、强化学习等，让机器人能够通过经验改进其行为。 9. **机器人设计和制造**：关于创造具有人性化外观和行为的机器人，以促进与人的亲和力。 10. **应用案例**：可能包含医疗、教育、娱乐、家庭服务等不同领域的社会机器人实例，展示社会机器人如何改变我们的生活。这本《Social Robotics》是了解该领域前沿理论和技术的重要资源，对于科研人员、工程师以及对社会机器人感兴趣的人来说，是一本宝贵的参考资料。通过阅读这本书，读者可以深入理解社会机器人学的最新进展，并可能找到新的研究方向或项目灵感。

XVI Table of Contents

Human Augmentation, Rehabilitation,

and Medical Robots I

Impedance Control of a Rehabilitation Robot for Interactive

Training ........................................................ 526

Wei He, Shuzhi Sam Ge, Yanan Li, Eﬃe Chew, and Yee Sien Ng

Design and Development of a Wearable Rehabilitation Robot .......... 536

Wei He, Shuzhi Sam Ge, Weian Guo, Zhen Zhao, Jie Zhang,

Shengtao Xiao, and Fon Ping Quek Nuraisha

Development and Path Planning of a Biped Robot .. ...... ...... ..... 546

Rui Li and Zhijun Li

Modelling and Trajectory Planning for a Four Legged Walking Robot

with High Payload .. ........ ...... ...... ........ ...... ...... ..... 552

Lorenzo Gagliardini, Xinghua Tian, Feng Gao, Chenkun Qi,

Christine Chevallereau, and Xianchao Zhao

Combining Gait Research of the Quadruped/Biped Reconﬁgurable

Walking Chair with Parallel Leg Mechanism .. ...... ...... ...... ..... 562

Xing Hu, Hongbo Wang, Lingfeng Sang, Qifang Gu, and Lin Yuan

Design of an Automatic Rotatory Chair Prototype for BPPV

Treatment ...................................................... 572

Fei Xu, Dingguo Zhang, Xueguan Gao, and Shankai Yin

Modeling and Control of a Lower-Limb Rehabilitation Robot .......... 581

Yanjiao Ma, Wei He, and Shuzhi Sam Ge

Human Augmentation, Rehabilitation,

and Medical Robots II

Particle Swarm Optimization Based Design for Knee Joint of Wearable

Exoskeleton Robot .... ...... ........ ...... ...... ...... ....... .... 591

Jia-yuan Zhu and Hong Zhou

The Application of Machine-Learning on Lower Limb Motion Analysis

in Human Exoskeleton System ..................................... 600

Cao-yuan Zhao, Xiang-gang Zhang, and Qing Guo

Hydraulic Pressure Control System Simulation and Performance Test

of Lower Extremity Exoskeleton .. ........ ...... ...... ...... ....... 612

Qing Guo, Xiang-gang Zhang, Dan Jiang, and Lu-lu Zhang

Dynamic Characteristics Study of Human Exoskeleton Based on

Virtual Prototype ... ........ ...... ...... ........ ...... ...... ..... 621

Wen-ming Cheng, Fang Liu, and Jian-bing Shao

Manipulating Mental States

through Physical Action

Jesse Gray and Cynthia Breazeal

MIT Media Lab, 20 Ames Street, Cambridge, MA

{jg,cynthiab}@media.mit.edu

Abstract. We present our implementation of a self-as-simulator archi-

tecture for mental state manipulation through physical action. The robot

attempts to model how a human’s mental states are updated through

their visual perception of the world around them. This modeling, com-

bined with geometrically detailed, perspective correct simulations of the

immediate future, allows the robot to choose actions which inﬂuence the

human’s mental states through their visual perception. The system is

demonstrated in a competitive game scenario, where the robot attempts

to manipulate the mental states of an individual in order to win. We

evaluate people’s reaction to the system, focusing on the participants’

perception of a robot with mental state manipulation capabilities.

1 Introduction

This paper focuses on a demonstration of mental state manipulation in a com-

petitive game scenario and an evaluation of human reactions to this behavior.

The motivation for this work is to explore the connection between (hidden) men-

tal states of an embodied agent and the (observable and modiﬁable) world in

which they exist.

An embodied agent exists in the physical world, and though its mental states

are hidden, there are rich connections between the agent’s mental states and

the world in which it is operating. Observing an agent’s visual perspective and

physical actions can help inform a model of the underlying mental states caused

by those perceptions and causing those actions.

In order to modify the mental states of another agent, it is necessary to

manipulate the world such that the agent’s observations will update its internal

mental states to the desired conﬁguration. Usage of this critical skill encompasses

a very broad set of interactions - in a cooperative scenario we might call it

communication. In such a scenario we often collapse this concept into “tell” or

“show.” However the underlying goal remains: to change a mental state through

physical action on the world that will cause the agent’s perceptual system to

update their mental states in the desired manner. This can be demonstrated by

observing how we respond to failures of the exchange - e.g., by moving an object

into the line of sight of an observer who is not paying attention.

S.S. Ge et al. (Eds.): ICSR 2012, LNAI 7621, pp. 1–14, 2012.

 Springer-Verlag Berlin Heidelberg 2012

2J.GrayandC.Breazeal

Mental State Manipulation

Perceptual Modeling

(B)

(A)

Hide(Ball)

(C)

(A)

(B)

(C)

! Know(Human, Ball)

Hide(Ball)

Fig. 1. Systems such as BDI architectures (B) include ad-

vanced mental state manipulation but tend to operate in

simulation with highly abstracted actions and perceptual

models. Other systems (A) employ geometrically correct

perceptual models to infer mental states, however lack the

ability to proactively manipulate these states. The research

here (C) pushes into a new part of the space using detailed,

geometrically correct mental state modeling to form short

term plans for physical actions designed to manipulate the

mental state of humans.

In a competitive sce-

nario we call this same

ability deception. In

this case the world

is intentionally modi-

ﬁed (again taking into

account the perceptual

capabilities of the tar-

get agent) in such a

way that the observing

agent will form incor-

rect mental states.

We believe that for

a robot to robustly in-

teract with people, it is

important for the robot

to form goals in this

space of mental state

outcomes rather than

simply perform com-

municative actions as a

series of physical or auditory actions; this ability will allow the robot to succeed

in situations where unexpected obstacles prevent naive communication strategies

from succeeding.

Researchers have approached the modeling of mental states in many ways.

The research here focuses on short timescale (0 to 60 seconds), highly detailed

modeling of how the robot’s actions will aﬀect the human’s mental states through

the means of their perception. We attempt to embrace the physical performance

of action and examine the accompanying communication value, rather than try-

ing to abstract away action-performance. The robot’s goal is to plan a sequence

of actions to cause the human to believe what the robot wants them to believe

while also accomplishing its task objectives; depending on the scenario, this takes

the form of robust, (sometimes implicit) communication, or deception. This is

especially useful in contexts where information has value and can be revealed

by behavior. For example, in search & rescue, common ground can be implicitly

maintained by making sure others observe important actions (“room-searched”).

Another example is competitive foraging, where agents seek to acquire resources

without revealing their location. Section 5 describes related work in detail, how-

ever Figure 1 illustrates the area we are exploring with this work.

We present an implementation and a study of human reactions to a system

which allows a robot to take action in order to alter the mental states of a hu-

man according to the robot’s goals. The robot employs perspective taking as well

as self-as-simulator techniques to model the mental states of nearby humans. The

robot has a simple planner, however the goal space of the planner includes not

only the desired world-state, but also the mental states of the human. This allows

Manipulating Mental States through Physical Action 3

the robot to form and execute plans that include changes to the human’s mental

states. After describing a motivating demonstration scenario, we provide imple-

mentation details. The demonstration scenario is then used as part of a user study

to evaluate how people perceive a robot that has these capabilities.

2 Demonstration Scenario

Robot

Human

Robot's

Objects

Human's

Objects

Robot's

Goal

Human's

Goal

Dividers

Fig. 2. Top down view of the demonstration scenario,

a competitive game between the human and the robot.

The robot stays on the upper part of the diagram pic-

tured, and the human on the lower part. Each player

has access to a matching set of objects on the left side,

and each has their own goal area on the right side. The

game ends when each player has placed an object into

their goal - the robot wins if the two players placed dif-

ferent objects, the human wins if the objects are the

same. Occlusions block the view of each player from

the opposing player’s object and goal areas, however

they can see each other as they travel between the

object repository and goal.

The architecture for men-

tal state manipulation pre-

sented here makes heavy

reuse of the motor actions

and perceptual processing

that the robot uses for its

own behavior. As such, it

canapplytovariedcon-

texts, as long as the robot

is conﬁgured to perceive

and operate there. The sce-

nario chosen for this demon-

stration revolves around a

simple competitive game

played between the human

and the robot. The game is

illustrated in Figure 2.

The rules of this game

create a situation where

the player who goes sec-

ond has the advantage of

potentially seeing the item

played by their opponent. If

the human goes second and

sees the item played by the

robot, it is straightforward for them to win by playing the object they saw the

robot play.

For this demonstration, the robot takes its turn ﬁrst. It is thus to the robot’s

advantage to manage the information that can be observed from its behavior. If

it proceeds in a straightforward manner, the human will be able to watch and

observe the object the robot plays, then play the same object and win. To win,

the robot must instead hide this information from the human.

For the demonstration, the game was played three diﬀerent times, each time

with a diﬀerent set of mental state goals for the robot (see Figure 3). These

diﬀerent mental state goals change the behavior of the robot as it plays the

game. In each case, the robot has the same overall task goal - transport the

cylinder to the goal location. However, the way it accomplishes this task varies

in the three conditions based on the mental state goals.

剩余678页未读，继续阅读

我是来学习的a

粉丝: 0
资源: 3

社交机器人：2012年国际会议论文集

Robotic Dance in Social Robotics—A Taxonomy.

usr-lab.github.io：此存储库包含Uppsala Social Robotics Lab的所有文件

robotics system toolbox 和robotics toolbox for MATLAB异同

robotics system toolbox 和robotics toolbox for MATLAB异同点

Robotics System Toolbox用法

Robotics System Toolbox怎么安装

robotics system toolbox 和robotics toolbox for MATLAB具体描述二者的异同

robotics toolbox10.4下载

Autel Robotics（的开源无人机

pip安装Gym-Robotics

最新资源