A Virtual Prototyping System for Operator Visibility of Earth-moving Equipment
R. Ingram,* R. Coddington* (Agr. Engr.), C. S. Larson,* E. Zentmyer
Caterpillar Inc.

This project involved the integration of a prototyping system for operator visibility into a visual virtual reality system. Use was made of existing data and information about cab geometry to evaluate operator visbility. The software developed allows design changes to be simulated in order that the results may be visually reviewed and evaluated.

This study involves the implementation of neural networks to replace the operator of an earth-moving vehicle. The system will make use of existing simulation software that can control vehicle performance and can add to the operator input from the neural network process. The long-term goal is to eliminate the operator from the process, and this effort draws from previous efforts of other researchers who have had limited success in this regard.

Research to date has shown that dividing the rider-applied torque between both wheels of a bicycle has strong advantages. Research is underway to develop reliable torque transfer mechanisms to permit rider power to be applied to the front tire in addition to the rear tire. A second aspect of the research is to experimentally document the changes in bicycle performance. Factors being investigated include mechanical efficiency, changes in rolling resistance, maneuverability, open loop stability aspects, and inhibition of loss of control.

The objective of this study is to develop advanced methodology for modeling of earth-moving vehicles. The current software using DYNASTY, a special dynamic simulation package, is enhanced by the use of Kane's method, which greatly reduces computer time for simulation runs. In addition, attention is given to the use of neural networks in the representation of the force effects imposed by the soil in the digging process. The project has far-reaching financial implications relative to the monetary strategies of an earth-moving equipment manufacturer.

This study involves the development of a virtual reality vision system that can be used for driver-operator training as well as design of the vehicle itself. Special effort is being made to reduce the computational time in the simulation process to simulate the systems in real time. The process is being applied to several vehicles and the output is being expanded to include sound as a part of the simulation.

The objective of this study is to evaluate operator visibility of earth-moving equipment using Caterpillar's Virtual Prototyping System. This system uses real-time interactive graphics by means of the National Center for Supercomputing Applications CAVE ®. Equipment part designs are converted from CAD files and placed directly into the virtual environment, giving an accurate representation of the vehicle, which is controlled using a dynamics simulation package. Then, within the environment, qualitative and quantitative studies on operator visibility can be performed. By using this system first, operator visibility can be evaluated much more quickly and inexpensively than by making a physical prototype.

This project involves the development of a virtual prototyping system for a wheel loader. The project includes an operator platform, a vehicle simulation, and design and construction computer graphics for use in the prototyping of a loader in a virtual reality system. This prototyping system brings an operator into the loop and allows real-time interaction with the vehicle. Roading and operator activity can then be compared between many designs. Proper use of the virtual prototyping system will allow more designs to be tested and in less time than testing of real wheel loaders.

The interaction between a wheel loader bucket and the soil is a very complex and difficult process to model. Previous attempts have either used soil models that are not well-suited to this application or lengthy computational simulations that model the individual soil elements. This model identifies and applies the individual forces that act upon the bucket during the soil-tool interaction, using simple spring-damper components. These individual forces then comprise the total resistive force that a vehicle encounters as it enters a pile.

The objective of this project is to develop a system to calculate the optimal path from given starting and ending points for an earth-moving vehicle to follow during a typical work cycle. Constraints considered in calculating the optimal path are the vehicle geometry, vehicle performance limits, work area configuration, and vehicle jerk and acceleration limits. The applications of this research will include use as a design tool to assist engineers in determining vehicle specifications.

Rather than engaging in ``action science'' (Acker, 1989) by building tools to effect change, the TEAM (Team Engineering Analysis and Modeling) project focuses on the empirical analysis and modeling of communicative acts engaged in by engineering design team members. TEAM draws the expertise and perspectives from human communication (Contractor and O'Keefe), human-machine interaction (Jones), and engineering design (Case and Lu) to investigate how engineers communicate in cooperative design teams. Such an interdisciplinary investigation will lead to better understanding of the nature of engineering teams and help to set a foundation upon which computer-based support tools are developed.

This project aims to develop a methodology to support the design of competitive products for ease of use by the disabled. Recent research has shown that over half the population has some form of disability that directly effects its ability to use off-the-shelf products. Often simple considerations by the designer can ensure that the design caters to typical difficulties of use. The methodology aims to give the designers an analytical tool for evaluating and improving their design for users with a range of common disabilities.

Much work has been performed to develop finite elements for the analysis of plates. These theories, primarily based on the Kirchoff assumptions, contain certain inconsistencies. In this study, a plate theory consistent with three-dimensional elasticity is developed using the theory of internal constraints. The basic assumptions of the Kirchoff plate theory are enforced as internal constraints as are the appropriate constrained constitute equations. In this way, no inconsistencies are introduced. A three-dimensional, 8-noded brick element is developed and yields satisfactory results in the analysis of some simple plate problems. The applicability of the element to more complicated problems is being investigated.

Controversy exists over various rate algorithms for the elastoplastic analysis. The oscillatory response in simple shear had led researchers to investigate numerous stress rates for the hypoelastic material response models. It is felt that the problems encountered here are due to more fundamental inconsistencies in their formulations. An attempt is made to investigate these issues by delving into the fundamental aspects of the hypoelastic formulation. Our findings will be exemplified by solving the simple shear problem using the revised theory.

The goal of this project is to develop a robust mathematical link between solid modeling and design sensitivity analysis. Two of the traditional difficulties of shape sensitivity analysis and optimization have been the troublesome and lengthy tasks of model parameterization and the evaluation of grid sensitivities essential to shape sensitivity analysis. While solid modelers and automatic mesh generation techniques have provided solutions for model parameterization, a method is being developed to analytically derive grid sensitivities from the variational geometry of a feature-based modeler.

Sensitivity analysis and optimization techniques are used to design control systems for nonlinear plants. These plants usually preclude classical/modern control strategies because of their complex nonlinear behavior. A rigorous dynamic model of the system under consideration is derived and an open loop control law is determined minimizing the desired cost function through sensitivity analysis and optimization techniques. Unfortunately, open loop schemes never completely solve automatic control problems as they lack desirable features of feedback control such as disturbance rejection and lowered sensitivity to parameter variation. Hence, we seek to combine open and closed loop strategies in the overall control scheme.

A computer-aided design methodology is developed to automate the design of mechanisms. The approach uses efficiently computed design sensitivities of the generalized coordinates and reaction forces with respect to local joint positions. Sensitivities are combined with numerical optimization to optimally locate joint coordinates to minimize a cost function while satisfying constraints. These sensitivities are efficiently computed because they utilize the decomposed Jacobian from the kinematic analysis. Mathematica ® is used to expedite the analytical derivations. The analytical sensitivities are verified with finite-difference sensitivities. Mechanisms with known analytical solutions are optimized for verification purposes. Finally, the design of a wheel loader bucket mechanism is studied.

To correlate analysis results with experimental data, optimization algorithms use first-order sensitivity expressions to minimize error functions with respect to the unknown model parameters. Research is being performed to investigate the stability of this solution process. Second-order sensitivities will be derived for the nonlinear thermal conduction systems. Methods for determining stability of the inverse solution based on the condition of the Hessian matrix will be explored.

First-order sensitivity expressions are derived for nonlinear thermal systems in a Eulerian reference frame with respect to surface heat flux terms. The surface region over which the flux is applied is defined through a series of design parameters. Local shape functions in the finite element formulation are manipulated to obtain accurate sensitivity information. The method is applied to optimize laser annealing processes. The solution ensures that a desired material microstructure is obtained with minimal laser power.

First-order design sensitivity expressions will be derived and implemented for various polymer-processing problems. The generalized Hele-Shaw model for Newtonian and power law fluids will be used in the fully coupled thermal fluid flow formulation. Initially, steady isothermal sensitivities will be developed to optimize the polymer sheet extrusion process to ensure a uniform sheet exit velocity by varying the thickness distribution inside the extrusion die. The finite element method will be used in nonlinear numerical simulations. Extension of the method to injection and compression molding with other non-Newtonian behavior is expected.

Explicit design sensitivity analysis formulations for systems undergoing large elastoplastic deformations will be derived. A continuum formulation will be investigated first by applying the direct differentiation method to the continuum governing equations. Next, a discretized version both in space and time will be obtained for finite element implementation. The derived formulations will be used to optimize metal-forming operations.