Teleoperated and Automatic Nanomanipulation Systemsusing Atomic Force Microscope ProbesMetin SittiDept. of Mechanical Engineering and The Robotics Institute, Carnegie Mellon [email protected] as a new emerging area enablesto change, interact and control the nanoscale phe-nomenon precisely. In this paper, teleoperated andautomatic control strategies for Atomic Force Micro-scope (AFM) probe based nanomanipulation applica-tions are introduced. Teleoperated touching to siliconsurfaces at the nanoscale is realized using a scaled bi-lateral force-reflecting servo type teleoperation controlwith custom-made 1 d.o.f. haptic device and AFMsystem. 1-D nanoforce sensing on the operator’s fin-ger is achieved during vertical or horizontal motion ofthe AFM probe tip on surfaces. Then, automatic con-stant height and force control strategies are introducedfor pushing nanoparticles and indenting soft surfaces.14nm radius gold nanoparticles are successfully posi-tioned with few nanometers resolution, and wax sur-face is indented with a conical AFM tip. Finally, ma-jor nanoscale control challenges are reported.Keywords: Nanorobotics, nanoscale systems and con-trol, nanomanipulation, Atomic Force Microscopy.1 IntroductionNanotechnology which aims at the ideal minia-turization of devices and machines down to atomicand molecular sizes has been a recent hot topic asa promising key technologies in this century. How-ever, for novel nanotechnology products, still thereare many challenges to be solved in the nano world,and nanomanipulation is one of the key challenges [1].Nanomanipulation is defined as the manipulation ofnanometer size objects with a nanometer size end-effector with (sub)nanometer precision. By manip-ulation, it is meant that nanoobjects are pushed orpulled, cut, picked and placed, positioned, oriented,assembled, indented, bent, twisted, etc. by control-ling external forces with sensory feedback. A basicnanomanipulation system structure is illustrated inFigure 1.This paper is focused mainly on developing con-trol strategies for nanomanipulation systems [2]. MainN a n o m a n i p u l a t o rA c t u a t o r # 2A c t u a t o r # 1N a n o F o r c e sN a n o O b j e c t sM a n i p u l a t i o nT a s kF o r c e a n dN e a r - F i e l d S e n s o r sF a r - F i e l d S e n s o r sE n v i r o n m e n t C o n t r o l a n d N o i s e R e d u c t i o n C o n t r o l l e r H u m a n - M a c h i n e I n t e r f a c eFigure 1: Basic structure of a generic nanomanipulationsystem.manipulation tasks are limited to some mechanicaltasks such as pushing/pulling, indenting, and touch-ing. As the control approaches, teleoperated and semi-autonomous control strategies as shown in Figure 2are investigated. In the former approach, a humanoperator directly in the control-loop manipulates thenanoobjects by using a man-machine user interface.Here, the operator controls the nanorobot directlyor sends task commands to the nanorobot controller.Hollis et al. [3] used teleoperation control for tactilefeedback from the nano world for the first time. In[4], [5], [6] and [7], force feedback and 3D real-timeVirtual Reality graphics display interface are utilizedduring teleoperation. Direct teleoperation approachcan realize tasks requiring high-level intelligence andflexibility. However, it is slow, not precise, not ex-actly repeatable, and engaged in many complex andchallenging scale difference problems. On the otherhand, the task-oriented approach avoids these prob-lems by executing only the given tasks in a closed-loop autonomous control [8]. In the automatic con-trol approach, nanorobot has a closed-loop control us-ing sensory information without any user intervention.However, the automatic control in the nano world isstill very challenging due to the complexity of thenanoscale dynamics, no available real-time nanoscalevisual feedback in ambient conditions, changing anduncertain physical parameters and disturbances, andinsufficient models and intelligent strategies [9].N a n o r o b o t C o n t r o l l e rM a c r o W o r l d :V i s u a l a n d H a p t i c D i s p l a yN a n o W o r l dN a n o r o b o t a n d S e n s o r sT e l e o p e r a t i o n C o n t r o l l e rS e n s o r y I n f o r m a t i o n(a)M a c r o W o r l d : U s e r I n t e r f a c e P a n e lN a n o W o r l dN a n o r o b o t a n d S e n s o r sS e n s o r y I n f o r m a t i o nN a n o r o b o t C o n t r o l l e r(b)Figure 2: Nanomanipulation control approaches: (a)direct teleoperation control; (b) semi-autonomous (task-oriented) control.2 Direct Teleoperation ControlAtomic Force Microscope (AFM) probe with itssharp tip is replaced by our finger at the nanoscale.For feeling the scaled nanoforces on the human finger,a bilateral teleoperation control system shown in Fig-ure 3 is proposed. In this 1-D force-reflecting servotype system, the operator controls the slave AFMprobe z position while feeling the resulting scaled per-pendicular nanoscale forces in her/his finger using a bi-lateral controller. A 1 degree of freedom haptic device[4] is used as the master device. The operator pushesthe bar of the haptic device, and the applied operatorforce fm, measured by a strain gage full bridge sensor,moves the bar down to a position xmwhich is mea-sured by a position sensor. Then, the nanoprobe zposition xsis controlled using a proportional-integral(PI) controller so that it can track the scaled mas-ter device position αpxm. The new xsposition re-sults in a nanoforce of fson the surface, and scalednanoforce αffsand master force difference is used tocalculate the linear motor torque using a proportional-derivative (PD) controller. This torque enables theforce feedback in the operator finger.M ( s )xmfmO p e r a t o rxsfsN a n o W o r l d-+apS ( s )afKf-++-tmtsFigure 3: Scaled bilateral teleoperation control system.The ideal response of the controller is given as fol-lows [10]:xs→ αpxmfm→ αffs(1)at the steady state. Here αf> 0 and αp> 0 arethe constant force and position scaling factors respec-tively. As the controller, a force-reflecting servo typecontroller is selected such thatτm= −αffs− Kf(αffs− fm)τs= Kv(αp˙xm− ˙xs) + Kp(αpxm− xs) (2)where Kpand Kvare proportional and differentialcontrol coefficients, and Kfis the force error gain.Using the slave and master dynamics equations, andassuming a very high tip-surface interaction