by
Franklin Ender Lee
A thesis submitted in partial fulfillment of the requirements for the degree of
Master of Science in Mechanical Engineering
University of Washington
1990
Interactive Real-Time Computer Simulation of Forest Harvesting Equipment
by Frank Lee
A method for simulating the operation of a feller-buncher was developed. The approach used a computer graphics workstation to generate views into a forest model. The data structure of the forest model was designed to allow for tree manipulation. Geometric solid modelling was used to define feller-buncher articulated components. Interactive user control of the feller-buncher was performed in real-time. The graphical animation used hidden surface removal and light reflectance rendering. The accuracy of the simulation and quality of its human interface were evaluated.
ACKNOWLEDGEMENTS
I would like to thank Professor Fridley for his mentorship in my work. Special thanks to Professor Ganter for his support and tutelage in computer graphics. I would also like to thank Larry Markee and John Andrews at Weyerhaeuser for their assistance in the woods.
TABLE OF CONTENTS
Chapter 1: Introduction
Simulation
Forest Harvest Simulation
Chapter 2: Background
Feller-Buncher Simulation with 2D Animation
Forest Vehicle Maneuvering Simulation
Feller-Buncher Simulation with 3D Wire-Frame Animation
Interactive 3D Color Feller-Buncher Simulation
Chapter 3: Approach
Goals and Objectives
Computer Equipment
Chapter 4: Computer Graphics
Coordinate Systems
z-Buffering
Light Modelling
Chapter 5: Models
Terrain Model
Tree Model
User Control Input
Vehicle Definition
Articulated Motion
Chapter 6: Vehicle Simulation
Discrete Time Step Control
Feller-Buncher Algorithm
Vehicle Motion
Cutter Head Location
Intersection Computations
Transcript Recording and Playing
Chapter 7: Results
Fell and Bunch Operations
Frame Rate Performance
Projected Performance
In the Woods Observation
Other Machine Models
Goals Evaluation
Chapter 8: Recommendation
Future Features
Stereo Viewing
Sound
Tactile Feedback
Related Issues
List of References
Appendix A: Color Figures
Appendix B: Source Code of Simulation
Appendix C: Data File Formats
Pocket Material: Videotape of Forest Harvesting Simulation
INTRODUCTION
Simulation
Simulation is the modelling of real systems. There are three general types of simulation, the use of: a physical model, analogous devices, and mathematical relations. If one considers a mass, spring, and damper as the real system, examples of the general types of simulation are as follows. A scaled down physical model of the system may be built, having similar behavior characteristics such as identical resonant frequencies and damping ratios. An analogous electrical circuit may be connected with values of inductance, capacitance, and resistance chosen to model the real system. Lastly, second order differential equations may be used to predict the linear response of the system.
Computers are often used to perform numerical solutions of the equations used in a simulation or to do repetitive calculations and manipulations of data. In engineering design, computer simulation is beneficial in various ways. The effects of design changes may be evaluated, without building costly prototypes, and optimization of design parameters may be done quickly through computerized iterations. Computer simulation also provides a measure of safety when the real systems being modelled are hazardous in nature, such as aircraft flight simulation. Flight simulation is also an example of interactive simulation, where user inputs at various stages of the simulation affect the process being modelled. Real-time performance of a simulation is delineated in that the processing of the simulation relations and the presentation of output results occurs at the actual rate of the real process.
Forest Harvest Simulation
The interaction of man and machine and nature endure as an important issues in forest harvesting. As the efficient management of natural resources grows in importance, previous concerns are being amplified. Impact to wildlife, erosion of harvested land, soil damage by harvesting procedures, and tree renewal growth rates, are just some of these concerns. In this study, an approach to the computer simulation of a swing-to-bunch feller-buncher is presented. Figure 1.1 shows an example of this type of feller-buncher. The uses of this type of simulation are two fold. The first being an aid in the design process of forest harvesting machinery, and the second being a means to evaluate current and possible new mechanized harvest methods, assessing their efficiency and environmental impact.
Figure 1.1: Swing-to-bunch Feller-buncher.
Chapter 2
BACKGROUND
This study is an extension of previous work in feller-buncher simulation. The work has developed and advanced with new ideas as well as with the availability of improved computing resources. The following gives an overview of the work leading up to this study.
Feller-Buncher Simulation with Two Dimensional Animation
(Fridley, 1985[1])
An approach for studying the operation of feller-bunchers being used for forest thinning was designed. The objective was to evaluate how various operating parameters such as: boom reach, boom swing speed, and machine travel speed, affected thinning performance. The simulation was composed of two main stages. The first, called Geometric Path Simulation (GPS), generated a path description file, used to define a series of feller-buncher operations. GPS consisted of an operator using a scale machine model on top of a forest map and digitizing table to define machine operations. The key point of this stage was that the operating strategies used in the forest thinning operation were determined by a human operator. The interface that the operator had to the simulation consisted of the machine model and the forest stand map.
The next stage of the simulation was known as Operating Time Simulation (OTS). OTS had five components: stand data, machine model description, path description, computer simulation program, and graphical animation. From the first four components, the times of each operation were computed. Given a machine motion, the time required for the operation was determined by operating rates (boom speeds and machine travel rates) defined in the machine model. OTS produced two outputs. Statistical information of the simulation was provided in printed output form, and a graphical animation of the simulation was used to verify the path model. The animation used simplified representations of machine and trees to display a plan view of the simulation.
Forest Vehicle Maneuvering Simulation
(Dyer and Fridley, 1988[2])
In this work, a personal microcomputer was used in developing a simulation program to test forest vehicle maneuverability. Two vehicle types were modelled, skid steer and articulated steer. The program produced a plan view drawing of the vehicle with trees spread throughout the terrain. Operators were given tasks in which they had to maneuver around the trees. Driving time and distances were recorded using various vehicle types and tree densities. The relationships of these variables were used to measure maneuverability. The graphical animation proved to be borderline adequate in creating a real-time interface to the user. The motions were not smooth in all circumstances. Faster personal microcomputers offered the possibility of attaining enhanced performance.
Feller-Buncher Simulation with Three Dimensional Wire-Frame Animation
(Wyman, 1984[3])
In this Masters Thesis, a graphical extension to the feller-buncher simulations by Fridley et al. was developed. GPS description files and OTS calculations were generated as before, but the graphical output was enhanced. A three dimensional animation of the simulation output was developed. The work employed distributed graphics processing. Distributed graphics was the use of multiple computer systems. In this case, non-graphical processing was done on a host mini-computer, and graphical processing was performed on a specialized graphics system. The graphics system used in this study was the Evans and Sutherland PS-300.
Using this approach, three dimensional wire-frame animations of the OTS were generated. Representations of trees and of a feller-buncher model were created. Views into the forest could be changed to observe the feller-buncher simulation from various orientations. The animation provided additional feedback in the verification of the overall simulation. Machine interference with standing trees as well as with itself, were visually identified.
Interactive Three Dimensional Color Feller-Buncher Simulation
(Block, 1988[4])
Using a Silicon Graphics, high-speed, raster graphics workstation, a real-time, interactive harvest simulator was designed. Previous feller-buncher work had not been real-time because the computer modelling was done as a batch process rather than immediate. Computer hardware performance proved to be the limitation. With the Silicon Graphics system, an interactive approach to the harvest simulator was possible.
The simulator that was created used a "through the windshield" view to look out into a forest. Using computer input devices, buttons and dials, an operator could control the motions and operations of a travel-to-bunch feller-buncher. Changes and actions that resulted from the inputs were computed, and the view was updated to reflect the results of the computations.
Chapter 3
APPROACH
Goals and Objectives
The main objective of this study was to integrate the features of the original work by Fridley et. al., where GPS and OTS were used, into a real-time interactive simulation. Elements of the simulation were to incorporate the use of three dimensional models with three dimensional computer animation. The development of the simulation centered on three areas; the user interface, modelling algorithms for vehicle and forest, and computer animation. The initial goals and prescriptions for each area were as follows.
In GPS, the operator used a stand map and a scale model of the harvest machine as the human interface with the simulation; the animated output from OTS did not play a role in operator decisions. In this study, computer animation will be used the interface between the operator and the simulation. The operator will supply control input using a button box device, and computer animation will serve as feedback from the simulation to the operator. The user interface should be designed to emulate actual operating conditions as closely as possible.
The modelling algorithms for the simulation involve vehicle motion and tree manipulation. Vehicle motion in the simulation must consider true operating performance; control speeds of vehicle components should be consistent with actual feller-buncher performance. For correct motion it also will be necessary to detect for vehicle collision with standing trees and other obstacles.
The forest model should contain the following minimum amount of detail. The terrain of the forest must incorporate elevation data to model uneven land plots. Data structures for the trees in the forest should include statistical and positional information about each tree. These ought to include; tree type, tree height, tree diameters, stump location, cut stem length, and cut stem position. The design of the tree data structures should allow for tree manipulation such as; felling a tree, moving a cut tree to be bunched, and being able to pick up cut trees for further processing.
Animation quality is influenced and evaluated by two factors, detail and smooth motion. However, image detail and smooth motion are opposing features to incorporate into animation. Increased detail in a scene will increase the time to draw it, thereby reducing the frame rate of the animation and having an averse effect on the display of smooth motion. Given the available computing resource of this study a good median between animation detail and frame speed must be determined. Initial image prescriptions require that image detail incorporate hidden surface processing and light reflectance rendering. For smooth animation, a frame rate greater than 15 frames per second must be attained. Assuming that both of the above goals are attained, additional rendering features such as shadows, texture mapping onto surfaces, and anti-aliasing of edges would be aspired for.
A secondary, forward looking design objective of the study was to create a simulation approach that could be used to model similar types of articulated machinery. Examples of other possible machinery are found in construction equipment, such as back hoes, bulldozers, and excavators. Specific to concepts related to mechanized, forest harvesting, the simulation should be flexible enough for various machines to manipulate trees within the forest model. The goal being to investigate the interaction of harvest machines between each harvesting stage, from cutting trees with a feller-buncher, to on location transportation by grapple-skidders, to on site processing with delimbers, and finally to loading logs onto trucks for final tree removal. The driving purpose is to improve forest resource management with increased harvest efficiency and to gain a better understanding of the relationships between man and machine and nature.
In summary, the goals for the simulation in this study were:
1) Create a user interface as true to life as possible.
2) Develop an interactive feller-buncher simulation capable of cutting and bunching trees.
3) Devise algorithms for permitting machine travel over uneven terrain and for detecting machine collisions with obstacles.
4) Implement computer animation with the best possible rendering features at a speed of 15 frames per second or greater.
5) Examine the possibility of extending the simulation to other forest harvest machinery.
Computer Equipment
A Silicon Graphics Personal IRIS graphics workstation was used to create the feller-buncher simulation. The Personal IRIS (Integrated Raster Imaging System) is a high-performance, high-resolution, graphics computing system. Features of the system make it ideal for three-dimensional computer graphics. The main power of the IRIS lies in a set of VLSI chips known as the Geometry Engine or Geometry Pipeline. The Geometry Engine processes graphical points, vectors, polygons, and curves using algorithms implemented in hardware rather than software. Other graphics features allow for the modelling of light sources and for smooth shading. These graphics features provided the capability of producing realistic images.
The particular configuration of the Personal IRIS used in the study was as follows. The system was based on a RISC processor running at 20MHz with a floating point coprocessor. System memory size was 16 megabytes. The graphics subsystem included 56 bit planes, allowing for 24 bit true color display and 24 bit hidden surface z-buffering. Performance for this configuration of the Personal IRIS was rated at 1.6 million floating-point operations per second (MFLOPS), 16 million instructions per second (MIPS), and a display rate of up to 20,000 four sided, Gouraud shaded polygons per second.
[1] Fridley, J.L.,Garbini, J.L, Jorgensen, J.E. and Peters, P.A., "An Interactive Simulation for Studying the Design of Feller-Bunchers for Forest Thinning.", Transactions of the ASAE (Vol.28,No3,pp.680-686,1985).
[2] Dyer, W.R. and Fridley, J.L., "Simulation of a Forest Vehicle Maneuvering Among Obstacles", American Society of Agricultural Engineers, Paper No. 88-7525, 1988.
[3] Wyman, R.P., "Animated Output for Simulations with Application to Feller-Buncher Designs", Masters Thesis, Mechanical Engineering University of Washington, 1984.
[4] Block, W.A. and Fridley, J.L., "Computer Simulation of Forest Harvesting Machine Concepts", American Society of Agricultural Engineers, Paper No. 88-7523, 1988.
[5] Foley, J.D. and Van Dam, A., Fundamentals of Interactive Computer Graphics, Addison-Wesley Publishing Company, Massachusetts, 1983.
[6] Rogers, D.F., Procedural Elements for Computer Graphics, McGraw-Hill Book Company, New York, 1985.
[7] Geometric Modelling of Solids, software program by JML Research Inc., Madison Wisconsin.
[8] Fridley, J.L., "A Method To Examine Effects of Feller-Buncher Boom Design and Operational Parameters On Forest Thinning System Performance", Doctoral Dissertation, Mechanical Engineering University of Washington, 1984.
[9] Greene, W.D., Fridley, J.L., and Lanford, B.L., "Operator Variability in Interactive Simulations of Feller-Bunchers", Transactions of the ASAE (Vol.30,No.4,pp.918-921,931,1987).