This page contains the abstracts and introductions to several of my controls related design projects. A compressed file containing the full document can be found in most cases. The compression program used is PKZip 2.04, and you can get a copy of the decompression program, as well as instructions on how to decompress the files here.
Most documents were created in Microsoft Word 6.0, and will look best in that format. If you do not have Word 6.0 or a newer version available, the Rich Text Format should work with most word processors. If you would like to see a copy of a document, but are having trouble downloading it, please send me a message at: harriman@euler.berkeley.edu
This report describes the tasks completed to provide a working proof of concept model of telecontroller. Telerobots are robots that are controlled by following the arm motion of a human operator by means of a telecontroller. Also, telerobots simulate environmental interactions through force feedback to the telecontroller device. A computer provides a means of communication between the telerobot and telecontroller and controls the accuracy of the results. In our project, potentiometers and magnetic particle brakes were used to provide position control and force feedback, respectively, for a two-link telecontroller. The telerobot used in our application was a planar SCARA. At the completion of the project, accurate position control was achieved. The "feel" of the telerobot through the brakes on the telecontroller was also satisfactory. The brakes provided damping to prevent position errors between the telerobot and telecontroller due to the velocity limitations in the SCARA. Also, a simulation of a wall through the computer gave informative results. Future work is needed, possibly in noise elimination, to prevent a "sticking" problem when the operator attempts to withdraw from the wall. The team recommends, once available, that more amperage be used to utilize the potential of the brakes and that a force transducer be used to sense external collisions. This along with elimination of various system noise problems should further our verification of the feasibility of a telecontroller.
Embedded computing systems have become more and more prevalent in mechanical systems of all types. The Train-Robot system studied in this project utilizes embedded computing for the control of a complex mechanical system. System control using transition logic topography was developed to accomplish the task of simultaneous transference of two balls from any of four bins to any four bins via the train-gantry system and robot manipulator. The mechanical system consisted of two train-gantry systems between which a robot arm can move small objects. The control software incorporates two processors running in a multitasking real time environment. These processors each control mechanical sub-systems while communicating via a dedicated serial communications link. In addition one of the processes is responsible for communication with, and control of the robot manipulator, while the other processor handles the overall scheduling of the complete mechanical system. The developed system control software was able to successfully control all of the mechanical sub-systems, as well as properly sequence their respective actions, all in real time. Performance analysis verified that the balls could be successfully moved from a variety of the initial conditions to a variety of the final positions, and that the software did indeed meet its real time requirements.
Guided missiles are high performance nonlinear systems. They have unstable dynamics, thus they require high performance nonlinear controllers to achieve any performance goal. This paper presents a control problem and a system model for the design of a nonlinear missile controller. To solve this problem, a modification to the system is first suggested. After the presentation of simplifying assumptions, the design of two controllers, a multiple surface sliding mode controller and a input-output linearization controller, are presented. The resulting system performance for each is shown to be satisfactory, and the results are discussed. Finally, suggestions for the implementation of the control scheme are made.
The goal of this project is to design and build a semi-autonomous robot capable of following a human leader. Such a robot has many potential uses. The physical strain put on the body by any task that requires a human to push or pull a heavy cart could be greatly reduced by the use of a follower robot. In addition, a follower robot could replace small utility vehicles in isolated environments. Since walking is much less dangerous than driving, this could lead to decreased frequency of accidents in places such as warehouses and factories. Also, the robots could be made to follow one after another behind a human leader, forming large caravans that require no special hardware to form or break.
To enable the Follower to track a human leader without the leader having to wear any special electronics, it uses a panning sonar. The sonar allows the robot to position the leader in two dimensions. Once the leader has been found, the sonar can range the distance to the leader, while the direction it has panned provides a relative heading. After finding the leader, the robot moves towards him/her, while attempting to keep the distance at a preset value.
The prototype Follower is a very simple robot. The major components of the system fall in three major categories, electronic, electromechanical, and mechanical. When properly interconnected, these components form the mechatronic system that is the Follower. The following sections describe the components that make up the Follower, how they are connected, and the software that runs the system. After the design of the follower is discussed, the performance of the resulting system is presented, as well as some helpful hints for designing and building an autonomous robot.
The entire text, as well as color pictures and video can be found here.