Mechatronics of the Humanoid Robot ROMAN

Similar documents
Emotional Architecture for the Humanoid Robot Head ROMAN

The Humanoid Robot ARMAR: Design and Control

ROBOT ROMAN FOR INVESTIGATIONS OF HUMAN-ROBOT INTERACTION

DEVELOPMENT OF THE HUMANOID ROBOT HUBO-FX-1

Design and Control of the BUAA Four-Fingered Hand

Korea Humanoid Robot Projects

ZJUDancer Team Description Paper

Mechanical Design of Humanoid Robot Platform KHR-3 (KAIST Humanoid Robot - 3: HUBO) *

Concept for Behavior Generation for the Humanoid Robot Head ROMAN based on Habits of Interaction

Sensor system of a small biped entertainment robot

Group Robots Forming a Mechanical Structure - Development of slide motion mechanism and estimation of energy consumption of the structural formation -

Team Description Paper: HuroEvolution Humanoid Robot for Robocup 2010 Humanoid League

Technical Cognitive Systems

SELF STABILIZING PLATFORM

Mechanical Design of the Humanoid Robot Platform, HUBO

ZJUDancer Team Description Paper Humanoid Kid-Size League of Robocup 2014

Autonomous Stair Climbing Algorithm for a Small Four-Tracked Robot

Realization of Humanoid Robot Playing Golf

Design and Implementation of a Simplified Humanoid Robot with 8 DOF

Team Description Paper: HuroEvolution Humanoid Robot for Robocup 2014 Humanoid League

ROMEO Humanoid for Action and Communication. Rodolphe GELIN Aldebaran Robotics

DEVELOPMENT OF A HUMANOID ROBOT FOR EDUCATION AND OUTREACH. K. Kelly, D. B. MacManus, C. McGinn

Elements of Haptic Interfaces

ASPECTS ON THE DESIGN OF A TRACKED MINI ROBOT DESTINED FOR MILITARY ENGINEERING APPLICATIONS

A Semi-Minimalistic Approach to Humanoid Design

Speed Control of a Pneumatic Monopod using a Neural Network

ZJUDancer Team Description Paper Humanoid Kid-Size League of Robocup 2015

System Overview of The Humanoid Robot Blackmann

On Observer-based Passive Robust Impedance Control of a Robot Manipulator

Studuino Icon Programming Environment Guide

Kid-Size Humanoid Soccer Robot Design by TKU Team

Shuffle Traveling of Humanoid Robots

HAND-SHAPED INTERFACE FOR INTUITIVE HUMAN- ROBOT COMMUNICATION THROUGH HAPTIC MEDIA

SRV02-Series Rotary Experiment # 3. Ball & Beam. Student Handout

UKEMI: Falling Motion Control to Minimize Damage to Biped Humanoid Robot

Motion Control of a Three Active Wheeled Mobile Robot and Collision-Free Human Following Navigation in Outdoor Environment

Design and Development of Novel Two Axis Servo Control Mechanism

Development of a Walking Support Robot with Velocity-based Mechanical Safety Devices*

Structure Design and Analysis of a New Greeting Robot

RoboCup TDP Team ZSTT

HexGen HEX HL Hexapod Six-DOF Positioning System

AN HYBRID LOCOMOTION SERVICE ROBOT FOR INDOOR SCENARIOS 1

The control of the ball juggler

Biomimetic Design of Actuators, Sensors and Robots

Design of a Compliant and Force Sensing Hand for a Humanoid Robot

HexGen HEX HL Hexapod Six-DOF Positioning System

A Passive System Approach to Increase the Energy Efficiency in Walk Movements Based in a Realistic Simulation Environment

sin( x m cos( The position of the mass point D is specified by a set of state variables, (θ roll, θ pitch, r) related to the Cartesian coordinates by:

Structure Design of a Feeding Assistant Robot

ROBOTICS ENG YOUSEF A. SHATNAWI INTRODUCTION

The Production and Research for Humanoid Robot

Robo-Erectus Jr-2013 KidSize Team Description Paper.

Gesture Identification Using Sensors Future of Interaction with Smart Phones Mr. Pratik Parmar 1 1 Department of Computer engineering, CTIDS

MEM380 Applied Autonomous Robots I Winter Feedback Control USARSim

Chapter 1 Introduction

HexGen HEX HL Hexapod Six-DOF Positioning System

العطاء رقم )7106/67( الخاص بشراء أجهز لقسم الهندسة الكهربائية على حساب البحث العلمي

Robot: icub This humanoid helps us study the brain

Development of Humanoid Robot Platform KHR-2 (KAIST Humanoid Robot - 2)

1 st IFAC Conference on Mechatronic Systems - Mechatronics 2000, September 18-20, 2000, Darmstadt, Germany

Application Note # 5448

An In-pipe Robot with Multi-axial Differential Gear Mechanism

Mechatronics Educational Robots Robko PHOENIX

Pushing Methods for Working Six-Legged Robots Capable of Locomotion and Manipulation in Three Modes

Introduction to robotics. Md. Ferdous Alam, Lecturer, MEE, SUST

Team Description Paper: Darmstadt Dribblers & Hajime Team (KidSize) and Darmstadt Dribblers (TeenSize)

Complex Continuous Meaningful Humanoid Interaction: A Multi Sensory-Cue Based Approach

Advanced Distributed Architecture for a Small Biped Robot Control M. Albero, F. Blanes, G. Benet, J.E. Simó, J. Coronel

PICK AND PLACE HUMANOID ROBOT USING RASPBERRY PI AND ARDUINO FOR INDUSTRIAL APPLICATIONS

More Info at Open Access Database by S. Dutta and T. Schmidt

A NOVEL CONTROL SYSTEM FOR ROBOTIC DEVICES

Stability and Dynamics and this all on a Ball?

Perception. Read: AIMA Chapter 24 & Chapter HW#8 due today. Vision

Fiber Optic Device Manufacturing

New Skills: Finding visual cues for where characters hold their weight

Automatic Testing of Photonics Components

Active Vibration Isolation of an Unbalanced Machine Tool Spindle

Chapter 2 High Speed Machining

Shuguang Huang, Ph.D Research Assistant Professor Department of Mechanical Engineering Marquette University Milwaukee, WI

Small Occupancy Robotic Mechanisms for Endoscopic Surgery

Vertical and horizontal Turning/Grinding Centers

Masatoshi Ishikawa, Akio Namiki, Takashi Komuro, and Idaku Ishii

M-041 M-044 Tip/Tilt Stage

FLL Robot Design Workshop

The Use of Social Robot Ono in Robot Assisted Therapy

A Musculoskeletal Flexible-Spine Humanoid Kotaro Aiming at the Future in 15 years time

Prof. Ciro Natale. Francesco Castaldo Andrea Cirillo Pasquale Cirillo Umberto Ferrara Luigi Palmieri

IDEA Connections. User guide

AMG-GR Series Gear-Driven Gimbals

A Machine Tool Controller using Cascaded Servo Loops and Multiple Feedback Sensors per Axis

Sensors and Sensing Motors, Encoders and Motor Control

A Do-and-See Approach for Learning Mechatronics Concepts

Stationary Torque Replacement for Evaluation of Active Assistive Devices using Humanoid

Robotic System Simulation and Modeling Stefan Jörg Robotic and Mechatronic Center

Chapter 1 Introduction to Robotics

MAGNETIC LEVITATION SUSPENSION CONTROL SYSTEM FOR REACTION WHEEL

Design and Control of an Anthropomorphic Robotic Arm

Playing Tangram with a Humanoid Robot

Birth of An Intelligent Humanoid Robot in Singapore

Project Number: P13203

Transcription:

Mechatronics of the Humanoid Robot ROMAN Krzysztof Mianowski 1 and Norbert Schmitz and Karsten Berns 2 1 Institute of Aeronautics and Applied Mechanics, Faculty of Power and Aeronautical Engineering, Warsaw University of Technology, Nowowiejska str. 24, 00-665 Warsaw, Poland, kmianowski@meil.pw.edu.pl 2 Department of Computer Science, Robotics Group, University of Kaiserslautern, Gottlieb-Daimler-Strasse, 67663 Kaiserslautern, Germany, {nschmitz, berns}@informatik.uni-kl.de 1 Introduction Recent developments in the area of service robotics show an increasing interest in personal robots. Those personal robots can help to handle daily work and to entertain people. Both tasks require a robot that is able to communicate with people in a natural way. For these tasks non-verbal motions like gestures, mimic and body pose are more and more important. Several anthropomorphic robots like [1], [6], [8] and [7] have been build in recent years. Although several goals could be reached humanoid robots are not able to communicate with humans in a natural way. Our concept of non-verbal interaction is mainly based on on FACS [4], which consequently describes the motions of the skin, eyes and neck. The results of FACS are extended with information concerning body pose and the influence on man-machine communication [5]. Based on these research results we present the mechanical design and construction of upper body, eyes and neck for the humanoid robot ROMAN. The mechanical design is an extension of the previously build emotional humanoid head with 3DOF neck construction ([3], [2]). Several experiments with the previously build head revealed the importance of additional expressive motions to realize a more natural man-machine communication.

2 Krzysztof Mianowski and Norbert Schmitz and Karsten Berns 2 The Humanoid Robot ROMAN The mechanical design of the humanoid robot is adapted to Ekmans [4] and Givens [5] work. The possibility to express emotions is therefore mainly based on the availability of the so called action units which have to be combined to express a specific emotional state. Table 1 lists those action units which are very important and can be realized with the humanoid robot ROMAN. These motions include the movements of the artificial skin, the neck and the eyes. The movements of the upper body are not listed as action units so we decided to realize them similar to the natural human upper body motions. Figure 1(a) shows an image of the engineering construction of the humanoid robot while the silicon skin can be seen in Fig. 1(b). Table 1. List of all realized action units corresponding to Ekmans numbering system including the approximated maximum ranges of motion Skin Motions 1 Inner Brow Raise 1cm 2 Outer Brow Raise 1cm 9 Nose Wrinkle 1cm 12 Lip Corner Puller 1cm 15 Lip Corner Depressor 1cm 20 Lip Stretch 1cm 24 Lip Presser 1cm 26 Yaw Drop 10 Head Motions 51 Turn Left 60 52 Turn Right 60 53 Head Up 20 54 Head Down 20 55 Tilt Left 30 56 Tilt Right 30 57 Forward 2cm 58 Back 2cm Eye Motions 61 Eyes Left 30 62 Eyes Right 30 63 Eyes Up 40 64 Eyes Down 40 65 Walleye 30 66 Crosseye 30 Fig. 1. (a) Engineering drawing of the upper body, the neck and the head construction included in Zygote s 3D realistic human model (b) A previous version of the humanoid robot ROMAN (ROMAN = RObot human inter- Action machine) with human-like silicon skin

Mechatronics of the Humanoid Robot ROMAN 3 3 Design Concept and Construction Based on the insights of the previous chapter we present the mechanical design of upper body, the artificial eyes, the neck and the interior frame of the robot. 3.1 Upper Body The movements of the human spine will be approximated with three degrees of freedom in relation to the base in an open kinematic chain. The kinematic scheme of the lumbar spine is similar to the one used in the design of the neck, while ranges of motion should be appropriate to the functions of the spine. The ranges of motion include rotation over vertical axis (±60 ), inclination forward/backward in relation to horizontal axis ( 30 and + 40 ) and inclination left/right in frontal plane (±30 ). Figure 2(a) shows the engineering drawing of the upper body. It was assumed that two main boards will be located on the chest in the front and in the back side. For the synthesis and recognition of speech a four channel sound card and a loud-speaker in the front of the body will be applied. For the protection of the equipment the chest has to be adequately stable and safe, so the basic design of the chest has the form of a stiff box with artificial ribs for protection. The chest as a mechanical part should be adequately resistant to transfer gravitational and external moments/forces acting to the head and to upper limbs (arms, hands), and should be relatively lightweight. It was decided to use the typical shell-shaped design for the box with planar walls made as a bent of tick plates and welded duraluminium. The mechanism (see Fig. 2(a)) consists of the base, rotational fork 1 driven by electric motor with gear, special cross yoke bearing by ball bearing in the fork 1 for inclination forward-backward and fork 2 for inclination left-right side bearing by ball bearing in the cross yoke. Driving forces (torques) are generated by electric motors with gears chosen in such a way, that typical compact solution of the motor with planetary gear is mounted to the fork (1, 2) and each one propels by small toothed wheel the big one mounted to the cross yoke. The total mass of the upper body including arms and head is estimated with up to 50kg. To partially compensate the external torque over horizontal axes produced by gravitational forces, in the proposed solution it has been applied special additional bridges attached outside the cross yoke on each inclination axis with elastic elements for compensating changes

4 Krzysztof Mianowski and Norbert Schmitz and Karsten Berns of potential energy, i.e. external springs which can be seen in in Fig. 2(a). Motor Dimensioning M ext (α) = m g R sin(α) (1) F spring1 (α) = 2 k 1 (l 1 (α) l 1min ) (2) M spring1 = F spring1 d 1 (α) (3) M spring = M spring1 + M spring2 (4) M result = M ext + M spring (5) The dimensioning to the motors is mainly dependent on the necessary torque. Figure 3(a) shows a sketch of the forward-backward bending joint which will be used to calculate the necessary motor torque. The static external torque M ext (α) (see Eq. 1) is dependent on the mass of the humanoid robot. We assume an overall mass m = 50kg (including the upper body construction with about 15kg) at a distance R = 0.4m with a gravitational acceleration g = 9.81 m s 2. The force created by the backside springs is dependent on the spring constants k 1 = 1637.9 and l 1min = 0.113m as well as the current length of the spring l 1 (α) (see Eq. 2). The length l 1 (α) can easily be calculated with the help of geometry. The force F spring1 (α) is multiplied with two since two identical springs are mounted on each side. Equation 3 shows the generated torque which Fig. 2. (a) Engineering drawing of the upper body of the humanoid robot with integrated main boards, loudspeaker and the 3DOF hip with supporting springs (b) This image shows the assembled head and neck including the fourth neck joint (c) Mounted eye construction from the backside and (d) the frontal view with artificial skeleton

Mechatronics of the Humanoid Robot ROMAN 5 m R r 1 α h1 Fig. 3. (a) This sketch of the forward-backward bending motion is used to calculate the external and resulting torque in the rotational center (b) The upper diagram shows the external M external and the resulting torque M result and the lower diagram shows the torque generated by the backside (M spring1 ) and front side springs (M spring2 ) is only dependent on the generated force and the orthogonal distance d 1 (α) to the rotational center. Since the construction is not symmetrical a second torque generated by the springs on the front side of the construction has to be calculated in a similar way. The spring-generated torque M spring (see Eq. 4) can than be used to calculate the resulting torque M result as shown in Eq. 5. The maximum external torque of 126.11 kg m2 could be reduced with the help of the springs to 79.44 kg m2. s 2 s 2 Figure 3(b) shows the previous and the resulting torque as well as the spring-generated torques dependent on the angle α. From the analysis it is shown, that the driving system (motor with planetary gear and one step wheel gear) should generate about 60 65% of the maximum value of external gravitational torque. 3.2 Artificial Eyes The construction of the eyes must be compact and lightweight since space is limited and we must be able to move the eyeballs independently up/down and left/right. The upper eyelid has to be movable to assist the expression of emotions. Additionally a small camera has to be integrated in the eyeball which is connected to exterior electronic parts with a flexible cable. The eyeball shown in Fig. 2(c) slides in a spherical attachment and can be moved with 2 pins in the rear part of the eye. Each pin is directly

6 Krzysztof Mianowski and Norbert Schmitz and Karsten Berns connected to a gearwheel and a stepper motor to allow a high the positioning accuracy. The third gearwheel is used to move the upper eyelid. In comparison to the human eyeball which has a diameter of 23mm to 29mm the eyeball of the robot has a diameter of 46mm. We increased the dimension to include the lightweight camera (12g) and get an iris with a diameter of 16mm which is necessary for the camera. The complete construction of a single eye has a weight of 150g including the motors and gears. The eyeball alone has a weight of 38g and can be moved with a maximum torque of 126mNm (vertical axis) and 136mNm (horizontal axis). The eyes are able to move ±40 in horizontal and ±30 in vertical direction and can be moved from left to right or top to bottom in about 0.5s. The eyelid can be moved 70 up- and 10 downwards from the initial horizontal position while a blink of an eye can be realized in 0.4s. 3.3 Neck The formerly 3DOF design of the neck must be extended to a 4DOF design to realize motions like nodding which is important for non-verbal communication. The axis of the fourth joint is located next to the center of the head to realize a rotation along the heads pitch-axis. The connection between neck and fourth joint is build of aluminum which has an increased stiffness in contrast to POM (Polyoxymethylene). The design of this connection is shifted to the rear of the head to form some free space which is necessary for head rotation. The rotation is assisted by two springs with a maximum force of 27N each since the center of gravity is placed in the front part of the head. The motor is able to generate a torque of 1230mNm to move the head with all actuators and sensors. Figure 2(b) shows the construction of the fourth neck joint including gearwheel and motor. Additional information concerning the neck can be found in [2]. 4 Robot control architecture The actuator system of the robot consists of 21 different motors including seven dc-, six stepper- and eleven servo-motors.they are connected to the digital signal processor unit (circuit board) which consists of a DSP (Freescale 56F803) cooperating with a logic device (CPLD - Altera EPM70256). Additionally there are current supply elements, CAN-bus elements and amplifiers for the motors. This digital signal processing unit is able to preprocess different types of encoders and controlling up

Mechatronics of the Humanoid Robot ROMAN 7 to two dc motors, three stepper motors (via a TMC428 stepper motor controller) and six servo motors. Several DSP units are connected via CAN-bus to an embedded PC which is placed inside the robot case. Fig. 4. The controller structure for one joint The controller for dc-motor has been implemented as presented in Figure 4. The controller itself consists of a cascaded closed loop velocity controller and a closed loop position controller. The velocity controller is an advanced PI-Controller, where the proportional value for desired and actual inputs are separated. This type is called ReDus unit and has the advantage, that only via the proportional element frictions of the motors can be compensated and the desired input can be reached without having a difference to the actual input. The integral element is only responsible to compensate external forces to the joint. This concept makes the controller quicker and more stable. The superposed position controller is combined with an anticipation element. The known velocity of the position calculation is an direct input to the speed controller. Differences in the position must be compensated by the position controller. No difference between the position actual and desired input must be build up to get a higher desired speed. The drag distance between desired and actual position value is shorter. This type of controller supports different mounted positions of the velocity and position sensor. Certainly this works only, if the desired values from the topper modules are correct and are available continuously. The robot control on the personal computer is implemented with the help of the Modular Controller Architecture (MCA). MCA is a modular, network transparent and real-time capable C/C++ framework for controlling robots. MCA is conceptually based on modules, groups and edges between them, which allow the implementation of hierarchical structures. The tools mcagui and mcabrowser allow the in-

8 Krzysztof Mianowski and Norbert Schmitz and Karsten Berns teraction with the system allowing the user to control and observe the robot. 5 Conclusion and Outlook Based on psychological research results concerning the interaction between humans we designed an emotional expressive humanoid robot. Equipped with 24 degrees of freedom and a complex sensor system the mechanical design allows complex interactions with humans in a natural way. It is also obvious that the integration of human-like arms will have a great influence on the communication task, so these will be added in near future. Based on this mechanical design it is necessary to extend the current behavior-based controlling architecture to realize first natural dialogs. References 1. T. Asfour, K. Berns, and R. Dillmann. The humanoid robot armar: Design and control. In International Conference on Humanoid Robots (Humanoids 2000), 2000. 2. K. Berns, C. Hillenbrand, and K. Mianowski. The mechatonic design of a human-like robot head. In 16-th CISM-IFToMM Symposium on Robot Design, Dynamics, and Control (ROMANSY), 2006. 3. K. Berns and J. Hirth. Control of facial expressions of the humanoid robot head roman. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Beijing, China, October 9-15 2006. 4. P. Ekman, W.V. Friesen, and J.C. Hager. Facial Action Coding System (FACS) - Manual, 2002. 5. D. B. Givens. The Nonverbal Dictionary of Gestures, Signs and Body Language Cues. Center for Nonverbal Studies Press, 2005. 6. S. Hashimoto. Humanoid robots in waseda univeristy -hadaly-2 and wabian-. In IARP First International Workshop on Humanoid and Human Friendly Robotics, 1998. 7. I.-W. Park and J.-H. Oh. Mechanical design of humanoid robot platfrom khr-3(kaist humanoid robot-3: Hubo). In IEEE-RAS International Conference on Humanoid Robots (Humanoids), 2005. 8. M. Zecca, S. Roccella, M.C. Carrozza, G. Cappiello, and et. al. On the development of the emotion expression humanoid robot we-4rii with rch-1. In Proceeding of the 2004 IEEE-RAS International Conference on Humanoid Robots (Humanoids), pages 235 252, Los Angeles, California, USA, November 10-12 2004.