G. Honderd

Model reference adaptive control of a gantry crane scale model

A method of reference model decomposition, an extension of model reference adaptive control, is presented. The decomposition method can be regarded as a way of including knowledge about the structure and parameters of unmodeled dynamics in the adaptive system, making it possible to choose a lower order controller which is only equipped for the nominal process part. To illustrate the decomposition method, an adaptive controller for a scale model of a gantry crane is presented. Simplifying and linearizing the mathematical equations describing the crane yields a fourth-order model, of which the dynamics of the load swing take two. A standard adaptive control algorithm needs eight parameters, which in practice yields unacceptable behavior. The decomposition method allows the use of only two adjustable parameters, and real-time experiments showing practical results obtained with the method are described.<<ETX>>

A Solution of the Unit Commitment Problem Via Decomposition and Dynamic Programming

Each day power generating units have to be selected to realize a reliable production of electric energy with the fewest fuel costs. This paper proposes decomposition and dynamic programming as techniques for solving the unit commitment problem, a high- dimensional non-linear, mixed-integer optimization problem. Experiments indicate that the proposed methods locate in less time a better solution than many existing techniques.

Model reference adaptive control of a direct-drive DC motor

An adaptive time-optimal position controller for a direct-drive DC motor with a design based on the model reference adaptive approach is presented. The high-acceleration torque of DC motors with permanent magnets permits direct coupling of the load to the motor axis, avoiding the use of a transmission with its inherent disadvantages (such as backlash and friction). However, direct coupling induces a large sensitivity to load variations, so the desired response and the reference model in the adaptive controller are adjusted to the motor capabilities. This is achieved by estimating the load inertia by means of a least-squares method and adjusting the reference model accordingly. The controller is tested on a direct-drive motor, and the results are compared with those obtained with a fixed proportional-integral-derivative controller.<<ETX>>

A new approach to on-line path planning and generation for robots in non-static environments

Abstract A new method for on-line path planning and path generation for a robot, in a non-empty, changing, working space is presented. The methodology is divided into three parts. The first part is centered on creating a pseudo-target, called ghost, and making the gripper pursue it. The movement of the ghost must be planned in accordance with the state of the real target, the state of the gripper, and the environment. The second part discusses the movement of the gripper (and the whole arm) through the Advanced Proportional Navigation (APN) algorithm, which is based on the well-known (Standard) Proportional Navigation (SPN). The main difference between them is the inclusion of the acceleration of both pursuer (gripper) and evader (target). The third part determines the profile of velocity for the movement of the gripper. The new APN algorithm (second part) is the chief subject of this paper. A computationally simple algorithm is presented. Simulations using a planar RR-type robot showed that the error in position during coarse motion was less than 0.3% and less than 190 mathematical operations were required.

Wall-following control of a mobile robot

The wall-following control problem is characterized by moving the robot along a wall in a desired direction while maintaining a constant distance to that wall. From ultrasonic distance measurements the distance and the orientation of the robot with respect to the wall can be calculated. This is solved by the use of an observer: the distance and orientation are estimated using a robot model and corrected by sensor measurements. Since a wall may not be available continuously (e.g., an open door), the robot must be able to navigate on its dead-reckoning as well. The feedback controller has been set up in such a way that it can handle both the observer data and dead-reckoning data. The controller has been verified by means of experiments. The results show a good performance with an absolute error of a few millimeters from the desired distance to the wall.<<ETX>>

Following A Wall With A Mobile Robot Using Ultrasonic Sensors

In this paper the wall-following control of a mobile robot is discussed. In literature, mostly vision systems are used to solve this problem. Typically, these systems provide an image of the environment in front of the robot. With this information the next (local) part of the trajectory is planned. Here, a mobile robot is considered without a sensor system to perform such a ‘look-ahead’ strategy. The wall-following problem is Characterized by maintaining a constant distance to the wall, which should be possible using a distance measuring sensor only. Ultrasonic sensors are cheap and fairly simple to use in this case. The main problem is the calculation of the distance and the orientation of the robot with respect to the wall from the sensor data. This is solved by the use of an observer. The controller has been verified by means of experiments. The results show a good performance with emrs of a few millimetres from the desired distance to the wall.

State estimators (observers) for the on-line estimation of non-measurable process variables

Abstract Bioprocesss control is frequently frustrated by the unfeasibility of measuring key variables of the process or by the inexactitude of on-line measurements. One way of approaching this problem is to use on-line measurements that can be made, in conjunction with a computational model of the process, to obtain estimates for the values of less easily measurable parameters. This information can then be used to control the process. The combination of model and measurement is known as an observer. The measurements taken are used to correct the state estimates provided by the model. The art of using observers lies in the development of good models and appropriate corrections.

Motion control of a mobile robot

Hierarchical layers of planning and control are a common approach to the mobile robot motion control problem. In this paper most attention is given to the implementation of the lower layers in a bottom-up approach. The layers consisting of motor speed control and position control must ensure at least reliable manoeuvering to be able to test different strategies in the high-level motion control. The lowest layer of speed control depends heavily upon the robot configuration and hardware. Therefore, effort has been spent in the modelling of the servo amplifiers and robot driving behaviour. These models have been validated through experiment. The position control layer implemented in software makes it possible to accurately achieve both absolute and relative moves. Through a description of the top layers of supervising and route planning it is indicated how intelligence and autonomy can be included in the motion control system. Here, a top-down approach is used. An interfacing level, trajectory control, completes the motion control structure.

A fast and efficient algorithm for the computation of path constrained time-optimal motions

The work, described in this paper, is performed as part of a collision-free motion-planning system for two cooperating robots in an assembly cell. The robots are of different types, an anthropomorphic type and a scara type.

Obstacle avoidance for a mobile robot

Obstacle Avoidance for a Mobile Robot A notorious problem in mobile obstacle avoidance is the detection and avoidance of obstacles. This thesis evaluates several well-known methods for controlling the motion of a mobile robot in an unknown dynamic environment. One of these methods, the Global Dynamic Window Approach, is selected and, using a laser range finder as the only range sensor, the method is implemented and tested on a mobile robot platform, a Pioneer 2 from ActivMedia. The result showed that the method is indeed an effective way for detecting and avoiding obstacles in real-time, in out-door tests the robot has traversed obstacle courses at velocities up to 1.2 metres per second. The method however showed to have some drawbacks; and should be combined with a higher-level algorithm that directs the robot to the best path.