Copyright © IFAC Robot Control, Wroclaw, Poland, 2003 A REAL TIME FRAMEWORK FOR PC·BASED ROBOT CONTROL UNDER THE QNX4 OPERATING SYSTEM Francesco Jatta • Emanuele Carpanzano' Giovanni Legnani ••• Antonio Visioli •• • Istituto di Tecnologie Industriali e Automazione, Consiglio Nazionale delle Ricerche, Milano, Italy •• Dipartimento di Elettronica per I'Automazione, University of Brescia, Italy ••• Dipartimento di Meccanica, University of Brescia, Italy ELSEVIER IFAC PUBLICATIONS www.elsevier.comllocalelifac Abstract: Even if highly reliable and endowed with powerful development tools nowadays industrial robot controllers don't usually allow the user to operate at the low level required in order to experiment control strategies and algorithms and to specify trajectories at the joint level. Such needs are mandatory in the robot control R&D area. Recently, the wide availability of computing power and operating systems with real-time capabilities offered the chance to develop low-cost PC-based solutions to overcome these limits. In this paper we present the design of a modular architecture for PC-Based robot control and contextually evaluate the use of QNX4 as a operating system for control applications. The main features of the designed controller have been employed for the motion control of a 4 d.oJ. reconfigurable parallel robot. Copyright © 2003 IFAC Keywords: Robot Control, Real Time Systems, Open Control I. INTRODUCTION Research needs in the robot control area often impose the necessity to be able to experiment both control strategies and algorithms, specify arbitrary trajectories at the joint level, experiment new task specification languages, take advantage of extra-sensing capabili- ties by integrating additional transducers in the control loop and so on. On the other side, nowadays industrial robot con- trollers, even if highly reliable and endowed with pow- erful development tools don't usually allow the user to operate at lowest level required to fulfill the above mentioned requirements. Moreover, the wide availability of low cost comput- ing power, interface cards for data acquisition, and operating systems with real-time capabilities made, in the last decade, PC-based solutions for robot mo- 551 Fig. I. The parallel kinematics machine used to vali- date the controller. tion control feasible (Casadei et al., 1998; Costescu et al., 1998; L10yd and Hayward, 1992; Corke and Kirkham, 2(01).