Robotic Assembly

Robotic Assembly

- Research Background

• Because the assembly margin of a precision part is smaller than the position accuracy of robots, a robot based on position control cannot assemble precision parts. In order to solve this problem, an assembly strategy based on force control is suggested.

• The position control based robot assembly is advantageous in that the assembly is performed within a short time. However, due to position accuracy problems, assembly may fail and in severe cases, collision with the environment can damage the assembled parts or damage the robot. In order to avoid this problem, a state estimation algorithm is proposed which detects the failure of assembly and increases the success rate of assembly.

- Research Objectives

• Force control for an assembly task

• Sensor compensation to eliminate effect of end-effector

• An assembly task based on blind search

• A state estimation algorithm to monitor abnormal assembly conditions

- Research Output

• Round peg-in-hole (Assembly margin: 0.08mm)

• 김병상, 김영렬, 송재복, 손승우, 6축 머니퓰레이터를 이용한 임피던스 제어 기반의 원형 펙 조립 (Impedance Control based Peg-in-Hole Assembly using a 6 DOF Manipulator), 대한기계학회 논문집 A, 35권, 4호, pp. 347-352, 2011.04.

• Lee, D.H., Na, M.W., Kim, Y.L., Song, J.B., 2017, “Model based assembly state estimation algorithm for small components of an IT device”, Ubiquitous Robots and Ambient Intelligence, pp. 740~743

    * Control system of manipulator

• Control system based on PC (CPU:dual core 3.0GHz)

• Using an external timer for a real-time control (sampling time: 1 ms)

• Using a motion controller for a position control (AJINEXTEK, PCI-N804)

    * Force control for an assembly task

• Impedance control based on position control

• Task space: impedance control using an admittance filter

• Joint space: position control

                                     < Control system using impedance control >

    * Sensor compensation

• Load identification: estimation of inertia parameters (mass, mass center)

• Gravity compensation : compensation of gravity force of assembly parts

• Contact force extraction

                          < Sensor compensation >

    * Sensor compensation: Load identification

• Using F/T sensor data and manipulator pose data

• Data acquisition operating only joint 5 and 6

• Estimation error: below 5%

    * Sensor compensation: Gravity compensation

• Computation of gravity force from current manipulator pose

• Gravity compensation error: below 1N

                                       < Gravity compensation >

    * Assembly strategy based on force control

• Assembly strategy based on blind search

• Contact detection based on sensing force

• Hole detection using peg position

                   < Assembly strategy for peg-in-hole task >

    * Assembly strategy based on position control

• Mobile IT parts assembly: Tablet PC parts (frame, mainboard, battery)

• Position control based robot assembly using 6-DOFs articulated robot (DENSO)

• Assembly strategy 1: mainboard to fixed frame

• Assembly strategy 2: battery to fixed mainboard 

                                        < Experimental setup of tablet PC assembly >

    * Assembly state estimation algorithm

• Assembly model is constructed by robot’s pose data and F/T sensor data

• From real-time measured data, it can detect an abnormal assembly state

• Assembly state: point, line, surface contact

                                    < Assembly state estimation for tablet PC assembly >

< Assembly state estimation algorithm >