Facilities @ Robotics Laboratory

This page is currently under maintenance. Information below may not be accurate or inadequate.

The Robotics Laboratory, containing three industrial robot manipulators—ABB Irb-6 and Irb-2000 and Irb2400/S4C+— together with the Open Robot Control architecture developed at the Dept of Automatic Control serves as a good common experimental platform for research activities from many different departments and research groups.

Matlab/Simulink interfaces for down-loading and dynamically linking new control algorithms to the robot systems and the integration of external sensors such as e.g. force/torque sensors and stereo vision cameras, also allow a lot of student projects and master thesis projects to use the facilities in the RobotLab. Our new robot system Irb2400/S4C+ donated by ABB during the year 2000 is currently integrated into our laboratory. Modification of the controller structure is done in close corporation with ABB Robotics.

Experimental Environment: pendulum an mpg-movie

The experimental platform uses real industrial robots from ABB with the control systems modified to permit new and customized control principles. A powerful host computer with work stations attached in a way to permit the use of standard mathematical software  for analysis of real-time data of the goal system—e.g., sensor-based motion control and adaptive algorithms  for path tracking.  To support prototyping of task-level programming and  autonomous operation, the IGRIP off-line programming system from Delmia, Inc. has been connected via the network to the robots in two ways: There is one on-line connection where the trajectories are computed in ENVISION (earlier IGRIP-utilizing world models etc.) and sent to the robot. The other type of connection works with robot programs exported from the off-line system to the file system. A novel mapping of such programs to and  from representations suited for factory floor operation (by the robot operator without extensive programming skill) has also been implemented. Concerning feedback control, it is important in applied robot control research to use robot manipulators that are commonly used in the industry. Simplified or special purpose research robots do not have realistic dynamic properties. The mechanical design done by major robot manufacturers has been worked out considering many application, quality, and maintenance demands. Control of such robots are industrially more relevant. As an example, there are many applications today where direct drive (rigid) robots are (today) not appropriate.

The experimental robot control system is developed unique in the sense that it is based on modern robots commonly used in the industry, still maintaining important safety functions of the original system, but allowing the researcher full access to control and programming functions. An overview of the reconfigured Irb-2000 system is shown in Figure 2. The master computer in the VME computer is based on a M68040 microprocessor. Supervision and safety functions are implemented on a M68030 board, well  separated from the rest of the system to prevent damage of the robot. Digital Signal Processors (DSP) are  used for low-level control and filtering of sensor signals. Sensors requiring very high data bandwidths are connected directly to the DSP boards. An additional DSP board belongs to the force-torque sensor (JR3). A six  DOF joystick (DLR) can be connected to a serial port of the M68030 supervision computer for data transfer to memory accessible from the VME bus. The Irb-2000 is equipped with AC motors, and because we want to be able to implement all of the control layers, the interfacing to the S3 control system used has to be at a lower level compared to the Irb-6 interface. The system is simply cut at the drive unit interface, which means that also the AC motor current references must be computed in such a way that the desired torque is achieved. The AC motor control should be computed with a rate of at least 1 kHz (preferably 2 kHz). Otherwise the torque control will not be good enough, particularly for high speeds. This means that the requirements on computing power are quite severe. Our solution is to use Digital Signal Processors (DSP). The same type of resolver-to-digital conversion as for the Irb-6 interface is used, but with the accuracy of 14 bits per motor revolution and 8 revolution counting bits in hardware. Both hardware and software are prepared for a resolution of 16 bits per motor turn. Thus, joint angles are provided as absolute 24 bit values. Using commercially available R/D converters (RDC) with an internal analog velocity signal and a phase-locked loop makes it possible to get proper anti-alias filtering by tuning that loop. This is hardly possible with optical encoders or with other types of resolver measurement principles (a higher sampling frequency and the roll-off of the process then have to be used instead). The 24 bit position data for the motors can simply be differentiated to get the speed; that signal has been filtered in the analog phase locked loop. The sampling period will then, however, be quite crucial. The next generation of the sensor interface will therefore also provide the speed value from the RDC chip. The R/D conversion hardware is located on the robot which reduces the required length of the wires for analog signals to a minimum.

The serial communication line of the original ABB S3 system was for our purposes too dedicated and involved with the rest of the hardware. Therefore it was realized that new communication hardware and software needed to be developed. Such an interface between sensors, actuators, and control modules was designed and built with two purposes in mind:

  1. To connect the Irb-2000 robot and its reconfigured control system in a simple and efficient way allowing sampling rates of 4 kHz for the motion control.
  2. To serve as a testbed for research within decentralized control and real-time systems.

This has thus far been successful. An early description of the design was given in [23]. The system has been used in a case study on distributed real-time control [34], and it has been referred to [36] as a platform for verification of timing problems in decentralized real-time control [35].

Figure 1: The principle used for connecting the joint servos of the robot to the control system. A sensor (input) and actuator (output) node sharing the same address can be viewed as a motor node from the control side.

Sensor System for Object Detection and  Localization
We consider fast and robust algorithms for our sensor system which is based on ultrasonic principles. The objective is to solve problems of object identification, object localization and fault detection in real time. The methods and algorithms should work also in an environment with disturbances. Early references are Knoll 1 (1991) and Watanabe and Yoneyama 2 (1992).

  1. Knoll, A. (1991). Ultrasonic Holography Techniques for Localizing and Imaging Solid Objects. IEEE Trans. Robotics and Automation. Vol 7, No 4, August 1991.
  2. Watanabe, S., M. Yoneyama (1992). An Ultrasonic Visual Sensor for Three-Dimensional Object Recognition Using Neural

Application Study Arc Welding
An important objective for robot applications such as arc welding is to define objectives and rules on a higher abstraction level than currently used sequential and explicit programming languages and to allow the user to command desired subgoals of the task directly. Our objective is to develop goal-oriented task-level control functions that describe the operation to be interpreted and executed in real time.

Figure 2: Overview of the Irb-2000 part of the laboratory, excluding gripper control and vision hardware.


Systems and Control Issues Features of interest are

© Lund Institute of Technology 2004. Web Master.

Last update: Friday, 04-Jun-2004 15:37:58 CEST