Kaveh Ashenayi

Autonomous local path planning for a mobile robot using a genetic algorithm

This work presents results of our work in development of a genetic algorithm based path-planning algorithm for local obstacle avoidance (local feasible path) of a mobile robot in a given search space. The method tries to find not only a valid path but also an optimal one. The objectives are to minimize the length of the path and the number of turns. The proposed path-planning method allows a free movement of the robot in any direction so that the path-planner can handle complicated search spaces.

Genetic algorithms for autonomous robot navigation

Engineers and scientists use instrumentation and measurement equipment to obtain information for specific environments, such as temperature and pressure. This task can be performed manually using portable gauges. However, there are many instances in which this approach may be impractical; when gathering data from remote sites or from potentially hostile environments. In these applications, autonomous navigation methods allow a mobile robot to explore an environment independent of human presence or intervention. The mobile robot contains the measurement device and records the data then either transmits it or brings it back to the operator. Sensors are required for the robot to detect obstacles in the navigation environment, and machine intelligence is required for the robot to plan a path around these obstacles. The use of genetic algorithms is an example of machine intelligence applications to modern robot navigation. Genetic algorithms are heuristic optimization methods, which have mechanisms analogous to biological evolution. This article provides initial insight of autonomous navigation for mobile robots, a description of the sensors used to detect obstacles and a description of the genetic algorithms used for path planning.

A neural network-based tracking control system

An application of the backpropagation neural network to the tracking control of industrial drive systems is presented. The merits of the approach lie in the simplicity of the scheme and its practicality for real-time control. Feedback error trajectories, rather than desired and/or actual trajectories, are employed as inputs to the neural network tracking controller. It can follow any arbitrarily prescribed trajectory even when the desired trajectory is changed to that not used in the training. Simulation was performed to demonstrate the feasibility and effectiveness of the proposed scheme. >

Evolving A Diverse Collection of Robot Path Planning Problems

This study presents an evolutionary computation system that can generate grid robot path planning problems. An evolvable cellular representation that specifies how to build a PPP is used. Also presented is a technique for taxonomizing path planning problems so that the vast number of problems that can be generated with the evolutionary computation system can be subsequently winnowed into a collection of substantially different problems of specified size. In this study the most difficult path planning problems, according to three different criteria, are evolved and those results are used to demonstrated the taxonomic technique. The hardness criteria are (i) the minimum number of turns a robot must make, (ii) the minimum number of forward moves it must make, and (iii) the sum of these quantities. A dynamic programming algorithm is used to compute these quantities for a given path planning program. The technique can be generalized to find cases of a specified hardness. The size of the board and maximum number of obstacles used are transparently specifiable.

Neural net-based identification of cells expressing the p300 tumor-related antigen using fluorescence image analysis.

We report on preliminary investigations of the use of an image analysis system to perform preliminary algorithmic classification of images of fluorochrome-labeled cells followed by capture of gray-level images of potentially abnormal cells for analysis by a neural network. Cells were labeled with an antibody against a bladder cancer tumor-associated antigen, and the neural net was used to distinguish true-positive cells from negative cells, false-positive cells (autofluorescent or nonspecific labeling), and cell-sized artifacts. Gray-level cell images were digitized and processed for analysis by a feed-forward neural network using back-propagation. The network was trained and tested with two independent image sets. Various network configurations and activation functions were investigated, including a sinusoidal activation function. At high power, the network agreed completely with the human observers classification. At low power, a strong clustering of cells classified by the network with expert classification was seen, while the neural network showed roughly 75% concordance with the human observer. In addition, a set of four features extracted from raw cell images were investigated. The features were: shape factor, texture, area, and average pixel intensity. A network trained with these features performed better than one operating with gray-level images. We conclude that using neural networks to recognize and classify images captured by an image analysis microscope is feasible.

Implementation of a learning fuzzy controller

This article describes our efforts at designing and implementing a practical learning fuzzy controller using inexpensive hardware. The controller engages basic control concepts and system-independent learning rules to enable it to adapt in real time to unknown plants even when it starts with a vacuous initial control policy. The controller remains dormant when the plant is operating satisfactorily and autonomously initiates online adaptation in real time when adverse performance is observed. The Intel-8031-based hardware implementation is geared for extensibility, robustness, and fault tolerance. Limited plant-dependent information is incorporated to tailor the hardware to applications. The design produces learning rates exceeding 200 reinforcements per second. The controller thus is able to learn to control unknown plants in real time even while it is controlling them. Physical experiments indicate that the learning fuzzy controller can rapidly and effectively deal with variations in plant characteristics, compensate for wear and tear, and handle disturbances and noise. >

A comparison of neural network and fuzzy c-means methods in bladder cancer cell classification

We report the performances of cancer cell classification by using supervised and unsupervised learning techniques. A single hidden layer feedforward NN with error back-propagation training is adopted for supervised learning, and c-means clustering methods, fuzzy and nonfuzzy, are used for unsupervised learning. Network configurations with various activation functions, namely sigmoid, sinusoid and gaussian, are studied. A set of features, including cell size, average intensity, texture, shape factor and pgDNA are selected as the input for the network. These features, in particular the texture information, are shown to be very effective in capturing the discriminate information in cancer cells. It is found, based on the data from 467 cell images from six cases, the neural network approach achieves a classification rate of 96.9% while fuzzy c-means scores 76.5%.<<ETX>>

Implementation of an on-line adaptive fuzzy controller in low-end hardware

Abstract The implementation of an on-line adaptive fuzzy control algorithm in low-end hardware is described. The adaptive controller starts with a vacuous control policy, i.e. its initial control surface is derived from a vacuous rule base. The control surface is adapted autonomously in real time using control meta-knowledge and basic learning constructs. Limited system-dependent information is incorporated to tailor the Intel-8031-based hardware to specific applications. This information includes input and output signal magnitudes and time-scale information. Simulations and actual experiments indicate that the on-line adaptive controller is very robust and fault-tolerant. A relatively high learning rate enables the controller to learn to control plants in real time even while it is controlling them.

Motor speed regulation using neural networks

An investigation is conducted of the use of the back-propagation neural network for motion control and speed regulation in industrial servo systems. The goal is to build an intelligent controller or regulator which has a versatility equivalent to that possessed by a human operator. The advantages of neural nets lie in that they are flexible in terms of learning and collective processing capabilities. Simulation was performed to demonstrate the feasibility and effectiveness of the proposed scheme. Network performance as a function of the number of hidden units and the number of training samples is addressed.<<ETX>>

A reliable system design for nondeterministic adaptive controllers in small UAV autopilots

Despite the tremendous attention Unmanned Aerial Vehicles (UAVs) have received in recent years for applications in transportation, surveillance, agriculture, and search and rescue, as well as their possible enormous economic impact, UAVs are still banned from fully autonomous commercial flights. One of the main reasons for this is the safety of the flight. Traditionally, pilots control the aircraft when complex situations emerge that even advanced autopilots are not able to manage. Artificial Intelligence based methods and Adaptive Controllers have proven themselves to be efficient in scenarios with uncertainties; however, they also introduce another concern: nondeterminism. This research endeavors to find a solution on how such algorithms can be utilized with higher reliability. Our method is based on using an adaptive model to verify the performance of a control parameter - proposed by a nondeterministic adaptive controller or AI-based optimizer - before it is deployed on the physical platform. Furthermore, a backup mechanism is engaged to recover the drone in case of failure. A Neural Network is employed to model the aircraft, and a Genetic Algorithm is utilized to optimize the PID controller of a quadcopter. The initial experimental results from test flights indicate the feasibility of this method.