Copyright © IFAC Robot Control, Vienna, Austria, 1991 SIMULATION ENVIRONMENT FOR ROBOT CONTROL DESIGN B_ Lantos Deparrment of Process Control, Technical University of Budapest, H-//// Budapest, Muegyetem rakpart 9, Hungary Abstract - - Simulation tools can support robot control design and help to find the most critical parts of the control algorithms. The paper describes a simulation system to examine a class of control algorithms containing usual industrial robot control (posi- tion, revolution and current control), computed torque technics and decentralized PID controllers, resolved motion acceleration control, hybrid position and force control and fuzzy control. The simulation system consists of a PC/ AT-386, a multitransputer card, graphic monitor and plotter. The software is well structured and only the inverse kine- matics depends on the robot type. A separate program running on PC can generate the dynamic model of the robot in symbolic form for nonrecursive computations and has a source-format compatible output to the simulation system. The system to be simulated is decomposed into robot and actuators, control algorithm and sensors. All parts can be prescribed in files containing help by using standard editors. Saturations and time constants of the sensors can be modelled. The transients of select able signals can be stored in files and drawn on screen and plotter by using a separate program. Illustration examples for SCARA type robots are discussed in the paper . Key Words -Robot control; simulation environment; multitransputer support. 1. INTRODUCTION acceleration E) can be recursively computed: n.4 and ia, = niq + 0 i (AL,iri_Iliti_l) 4>i AT-I,i<l>i-I + iWi X iti_14i (2) Characteristic for advanced robot control algo- rithms is that the usual structure of position control (internal current control, central revolu- tion control and out er position control) should be given up and the driving torque has to be controlled directly. The control is usually a hi- erarchical one and can be divided into specifica- tion level (robot programing language, graphic simulation, path design), algorithmic level and torque control (power electronic) level. The pa- per describes a simulation system to examine a class of control algorithms containing usual in- dustrial and some advanced robot control meth- ods. Only open chain rigid robots have been concidered. n i (AL,,{n i- 1 - [Pi_I ,ixlri _d lidi _l) 2. THE ROBOT DYNAMIC MODEL The base for advanced robot control algorithms is the robot dynamic model which gives the rela- tion between driving torque (T), inertia matrix (H) and the centripetal, Coriolis and friction ef- fects (G4) and the gravitational part (D) : H(q)q + h(q, 4) = (1) = H(q)q + C(q, 4)4 + D(q) = T. Using the Denavit-Hartenberg parameters the position pi-I,i , the orientation A i- I.i, the par- tial angular velocity 'ti_1 and partial velocity 'd i _ 1 between neighbouring links can be deter - mined and the kinematical quantities (velocity v, acceleration a, angular velocity wand angular 435 0 i AL ,d0'-1 + 4>i-1 x Pi-I,i + +i-I Wi _ 1 x C-IWi-1 x P,_I,.)} + +2 i wi x id._ 1 4. - iti_ 1 4i x idi_Iqi. It is useful to determine the kinematical quan- tities to the center of mass i (!ci: (3) 0", 0 i + <1>, x i I! ci + Using the mass mi and inertia moment Kc; be- longing to a coordinate system having origin in the center of mass but axes parallel to the De- navit-Hartenberg frame Ki the quantities in the dynamic model can be expressed similarly to Vukobratovic and Potkonjak (1985) using the recurrence formula TO,i-1 Po.i-{] n H + r; Kc.f.} H;[Kc.4>. - (Kc:w.l x ·w.]} G; ( R.; ) (4) M,