Cecil O. Alford

Coordinated control of two robot arms

A computer control structure has been designed and implemented to enable two robot arms to function in a coordinated manner. One algorithm has been derived and implemented but not tested for a closed-loop configuration. Errors for the open-loop operation are given in the report.

Applications of random restart to genetic algorithms

In this paper, a new genetic algorithm is introduced in which the mutation operation has been replaced with random restart. The new genetic algorithm is applied to the problem of scheduling a set of tasks onto a multiprocessor system. This problem is known to be NP-complete. Using the Markov chain method, the expected time for the algorithm to reach an optimal solution is derived analytically and computed for several different problem sizes. These results indicate that the expected time to reach the goal for the mapping problem increases sublogarithmically as the size of the search spaces grows exponentially. It is also shown that the expected time to reach an optimal solution is proportional to the time the algorithm spends among the transient states.

Computation of Orientational Filters for Real-Time Computer Vision Problems I

Orientational filters have been used frequently for computer vision problems. Despite their strength in various vision problems, their use in real-time applications has been limited since they are computationally intensive. This paper examines various types of orientational filters and their implementation schemes, and proposes an efficient computation method called separable approximation . There are three different algorithms for the approximation: Singular Value Decomposition, Orthogonal Sequence Decomposition and Singular Value/Orthogonal Sequence Decomposition. Implementation is described and evaluated based on throughput, latency, computational complexity and the amount of storage required. Performance of these algorithms is examined on Gabor filters, and their effectiveness is demonstrated.

Computation of Orientational Filters for Real-time Computer Vision Problems III: Steerable System and VLSI Architecture

Abstract Orientational filters have been used frequently for computer vision problems. Despite their strength in various vision problems, their use has been limited in real-time applications since they are computationally intensive. Part I presented separable approximation as a way to implement a real-time orientational filter operation with a small amount of hardware. Part II presented an efficient computation scheme for 2D multi-resolution decomposition. This paper extends the results of the previous papers to construct a multi-resolution filter system where the orientation of each filter can be adaptively controlled. Such a system is called steerable and is useful in many image analysis/processing applications. This paper also presents a VLSI architecture for various orientational filter systems. The architecture is scalable in terms of the input image size, the filter size, the number of orientational filters, and the approximation order.

Pascal’s Triangle: Top Gun or Just One of the Gang?

Pascal’s triangle can appear as a member of classes of triangular arrays where presumably no class member should be ranked in importance over any other. Two such cases which came to mind were the multinomial triangles [6] and the Hoggatt triangles [2]. No doubt there are others. We selected the multinomial triangles. Was Pascal’s triangle only a binomial triangle in a sea of trinomial, quadrinomial, pentanomial, etc., triangles, or might it exhibit a significant influence on the makeup of the other multinomial triangles? We admit a certain prejudice in our choice. Computer experimentation with partition counting, large multinomial expansions, and generating functions using computer algebra systems (muMath, Derive, Mathematica) hinted at a definite Pascal influence. A few years ago, such experimentation would have been virtually impossible.

Solving for manipulator joint rates in singular positions

Computing manipulator joint velocities and accelerations, given the end-effector Cartesian rates (velocities and accelerations), is complicated by the singular positions the manipulator can assume along a trajectory. At such configurations the Jacobi matrix that relates Cartesian end-effector rates to joint rates cannot be inverted. Geometric approaches in solving for joint rates yield expressions that are indeterminate at such configurations. It is shown that exact solutions for the joint rates can be obtained in most singularity situations if the system of differential kinematic relations is properly partitioned. A solution methodology is developed and illustrated by an example.<<ETX>>

Artificial neural network control of a nonminimum phase, single-flexible-link

A single-link flexible manipulator with a rotary actuator at one end and a mass at the other is modeled using the Lagrangian method coupled with an assumed modes vibration model. A SIMO state space model is developed by linearizing the equations of motion and simplified by neglecting natural damping. Laplace domain pole-zero plots between torque input and tip position show nonminimum phase behavior. Nonminimum phase behavior causes difficulty for both conventional and artificial neural network (ANN) inverse-model control. The most promising ANN method for the control of flexible manipulators does not appear to converge to a solution when the system is lightly damped. To overcome this limitation, a modified cost function is proposed. Simulations show that the ANN is able to converge to a solution even in the case of no damping. The modified approach fails, however, for beams exceeding some critical length measure. Identification of the critical length and proposals for extending the result are discussed.

Computation of Orientational Filters For Real-Time Computer Vision Problems II: Multi-resolution Image Decomposition

Abstract Orientational filters have been used frequently for computer vision problems. Despite their strength in various vision problems, their use has been limited in real-time applications since they are computationally intensive. A previous paper presented separable approximation as a way to implement a real-time orientational filter operation with a small amount of hardware. This paper extends the result of the previous paper to a multi-resolution decomposition.

A vision system with real-time feature extractor and relaxation network

This paper reviews our ongoing efforts on designing a real-time vision system. The system consists of a real-time feature extractor and a relaxation network. The filter system is capable of performing multiple 2D non-separable filters, multi-resolution decomposition, and steerable transform. The relaxation network is capable of performing various relaxation, diffusion and minimization operations. The paper shows several vision tasks which can be implemented effectively on the vision system.

The teaching of VHDL in computer architecture

There are problems in incorporating VHDL into the undergraduate curriculums beginning computer architecture courses. The problems relate mainly to cost arising from two factors: VHDL tool availability and proper lecture material to coincide with the course objective(s). At the Georgia Institute of Technology, pilot VHDL lecture materials have been developed to address these two issues.