W. Thomas Miller

Application of a general learning algorithm to the control of robotic manipulators

In this paper, we discuss the use of a general learning algo rithm for the dynamic control of robot manipulators. Unlike some other learning control schemes, learning is based solely on observations of the input-output relationship of the system being controlled and is independent of control objectives. Information learned previously can be applied to new control objectives as long as similar regions of the system state space are involved. The control scheme requires no a priori knowl edge of the robot dynamics and is easy to apply to a particu lar control problem or to modify to accommodate changes in the physical system. The control scheme is computationally efficient and well suited to fixed-point implementation. The learning controller is evaluated in a series of computer simu lations involving a two-axis-articulated robot arm during simulated repetitive and nonrepetitive movements. We inves tigate the effects of varying learning algorithm parameters as well as control system performance in the presence of obser vation noise and changing manipulator payloads. The learn ing control system presented promises to provide good dy namic performance in complex situations at a reasonable cost as measured in terms of both hardware and software devel opment.

Effect of High-Frequency Current on Nerve and Muscle Tissue

The stimulus threshold for nerve and muscle tissue as a function of frequency for sinusoidal electrosurgical current below 1 MHz was shown to be a monotonically increasing value. There was no frequency below tissue destruction threshold values where excitable tissue cannot be stimulated if the cufrent intensity is great enough. This points out misconceptions concerning electrosurgical maneuvers prevalent since the early 20th century, and may help in understanding potential surgical hazard conditions.

A Nonlinear Learning Controller for Robotic Manipulators

A practical learning control system is described which is applicable to complex robotic systems involving multiple feedback sensors and multiple command variables during both repetitive and nonrepetitive operations. The learning algorithm utilizes the Cerebellar Model Arithmetic Computer (CMAC) neural model developed by Albus. In the controller, the learning algorithm is used to learn to reproduce the nonlinear relationship between the sensor outputs and the system command variables over particular regions of the system state space. The learned information is then used to predict the command signals required to produce desired changes in the sensor outputs. The learning controller requires no a priori knowledge of the relationships between the sensor outputs and the command variables. The results of learning experiments using a General Electric P-5 manipulator interfaced to a VAX-11/730 computer are presented. These experiments involved learning to use video image feedback to track three dimensional task trajectories relative to objects moving on a conveyor. No a priori knowledge of the robot kinematics or of the conveyor speed or orientation relative to the robot was assumed. In all experiments, control system tracking error was found to converge after a few trials to within error limits defined by the resolution of the sensor feedback data.

Pattern Recognition Using A CMAC Based Learning System

This paper presents a new approach to image feature vector classification based on the Cerebellar Model Arithmetic Computer (CMAC) neural network proposed by Albus. This approach promises advantages both over traditional methods for feature vector classification and over other neural network based classifiers. One advantage is that the generalization properties inherent in the network allow the formation of highly nonlinear decision boundaries, and allow multiple disjoint regions of feature space to be defined in the same class. A second advantage is that the computation time required for network training and for vector classification is greatly reduced relative to other nonlinear classification techniques. Results from several classification experiments are presented, including the investigation of the effects of noise on classifier performance, and the learning of rotational classification invariance using feature vectors deliberately chosen to be highly sensitive to object rotation. Capabilities and limitations of this method of feature vector classification are discussed.

User interfaces for first responder vehicles: views from practitioners, industry, and academia

By the nature of their jobs first responders have to interact with in-vehicle devices even as they drive under challenging road conditions. In this paper we assess the state-of-the-art in creating safe in-vehicle user interfaces for first responders, and we propose six research and development priorities for future work in this realm.

Using speech, GUIs and buttons in police vehicles: field data on user preferences for the Project54 system

The Project54 mobile system for law enforcement developed at the University of New Hampshire integrates the control of disparate law enforcement devices such as radar, VHF radio, video, and emergency lights and siren. In addition it provides access to state and national law enforcement databases via wireless data queries. Officers using Project54 are free to inter-mix three different user interface modes: the device native controls; an LCD touchscreen with keyboard and mouse; and voice commands with voice feedback. The Project54 system was utilized by the New Hampshire State Police agency wide for a period of seven years spanning 2005 through 2011. This paper presents an analysis of user preferences in regard to user interface modes during the three years 2009 through 2011, obtained through logs of daily system use in approximately 200 police cruisers. Results indicate that most officers chose to use the touch screen controls frequently instead of the device native controls, but only a minority chose to use the speech command interface.

Practical Demonstration Of A Learning Control System For A Five-Axis Industrial Robot

The overall complexity of many robotic control problems, and the ideal of a truly general robotic control system, have led to much discussion of the use of neural networks in robot control. This paper discusses a learning control technique which uses an extension of the CMAC network developed by Albus, and presents the results of real time control experiments which involved learning the dynamics of a 5 axis industrial robot (General Electric P-5) during high speed movements. During each control cycle, a training scheme was used to adjust the weights in the network in order to form an approximate dynamic model of the robot in appropriate regions of the control space. Simultaneously, the network was used during each control cycle to predict the actuator drives required to follow a desired trajectory, and these drives were used as feedforward terms in parallel to a fixed gain linear feedback controller. Trajectory tracking errors were found to converge to low values within a few training trials, and to be relatively insensitive to the choice of feedback control system gains.

A low-cost masquerade and replay attack detection method for CAN in automobiles

Controller Area Network (CAN) is the main bus that connects Electronic Control Units(ECUs) in automobiles. The CAN protocol has been revised over the years to improve vehicle safety but the security of communication over a CAN bus is still a concern. Despite different kinds of attacks challenge the CAN security, the attack that injects masqueraded CAN frames is extremely difficult to defeat given the limited resources available in CAN system. We propose a low-cost detection mechanism to address the masquerade and replay attacks on the CAN bus. Existing work either requires to store a long list of legal CAN IDs or uses hardware-consuming cryptographic algorithms to detect attacks. In contrast, our method only adds one more CAN ID to the acceptance filter of the CAN node under protection, eliminating the need for cryptographic modules and significantly reducing the hardware cost. We implemented our method in a CAN system prototype. Our experimental results show that the latency overhead of the proposed method is approximately three orders of magnitude less than that of other methods. Our method is capable of detecting the masqueraded and replayed CAN frames with a detection speed of 40μs, which satisfies the real-time requirement of automobiles.

CLW 2014: The Fourth Workshop on Cognitive Load and In-Vehicle Human-Machine Interaction

Interactions with in-vehicle electronic devices can interfere with the primary task of driving and increase crash risk. Interactions with in-vehicle interfaces draw upon visual, manipulative and cognitive resources, with this workshop focusing on cognitive resources for which measurement processes are less well known or established. This workshop will focus on two methods of measuring cognitive load, the Decision Response Time Task and collecting eye fixation data. The workshop will describe and demonstrate how they are collected, and discuss how the resulting data are reduced and analyzed. The focus will be on practical aspects of collecting and analyzing data using these methods, not on reporting research results.

APCO Project 25 wireless data services for smaller public safety agencies

Digital data messages are very important in modern communication systems and the advanced data technologies that rely on the Internet Protocol have opened the door to a wide range of IP-based applications and services. With this trend in mind, we have implemented a system that will allow smaller law enforcement agencies to enable data services in their cruisers in a cost effective way, with a high level of reliability. Towards this goal, we have implemented an inexpensive Software Defined APCO Project 25 Data Base Station that utilizes the standard IP network interface. This paper describes the overall design of the data base station, its implementation in the preexisting radio network infrastructure, and the testing process. Laboratory tests have produced promising and encouraging results prior to real world deployment.