Proposal for Insect Video Tracking Device

Proposal for Insect Video Tracking Device

Project Number (Project 99.05)

Design Team Members:

George H. Sapna III

(302) 837 8817

219 Sussex Hall, Newark, DE 19717

Raymond Foulk

(302) 738 2884

18 Madison Drive, Newark, DE 19711

Ryan McDonough

(302) 369 9113

329 East Main Street, Apartment 210, Newark, DE 19711

Justin Combs

(302) 737 7281

29 East Park Place, Newark, DE 19711

Sponsor:

Keith Hopper

(302) 731 7330

501 South Chapel Street, Newark, DE 19711

Introduction

Pest control is an important issue in the agricultural industry. A significant amount of crops are damaged or lost each year due to harmful insects. Diuraphis noxia, a crop eating insect, is particularly a major problem for the United States wheat industry. Fortunately, these pests can be effectively controlled by the introduction of natural predators, such as the Aphelinus asychas. One problem with this method of pest control is that of population establishment. Therefore, it is necessary to understand the reproductive habits of beneficial insects if they are to be used successfully in controlling destructive insects.

Research is currently being conducted on the reproductive habits of the Aphelinusasychas species. One study is focused on the mechanism by which the males of the species are attracted to the females. The female A. asychas deposits a sex pheromone trail wherever she walks. The purpose of this is to lure the male A. asychas so that mating can take place.

Simple tests are being performed to study the male response to this pheromone. The male’s walking patterns are compared to those of a female’s, which had earlier traversed the same area. The need to follow the motion of the insects is the origin of our problem.

Background

Currently the insect movements are being observed using six 4-centimeter diameter arenas. In each arena, a female deposits the sex pheromone as it walks on the glass floor. Then, the female A. asychas is removed and a male A. asychas is introduced to the arena. The male movements are compared to the female movements using a fixed CCD camera coupled with a computer. A record of the insect movements is written to a data file as a time history of two-dimensional space coordinates. In order to accomplish this, a computer program must digitize the CCD analog image in real time, distinguish between the insects and the background, and compute the positions of the insects relative to their starting positions as sets of x and y coordinates. Records of the position coordinates are stored in files, which are later used in a program that computes the discrete speeds and the average speeds of all six insects simultaneously. The disadvantage of this system is that the resolution of the camera prohibits the tracking of these insects over much larger areas. It has been found that the 4-centimeter diameter arenas confine the insects, which alters their otherwise uninhibited walking behaviors. By broadening the arena domain, the insects’ movements will better resemble their natural behavior.

Description of Problem

The primary customer, Keith Hopper, has expressed a strong desire for a system that allows the insects to roam more freely in a larger arena. He would like to use the existing camera, which means that the camera will have to move in accordance with the insect’s movements due to the resolution problem discussed above. In order to follow the insect, the insect’s position must be identified at discrete time intervals, and a “smart” program must decide how to actuate motors that move the camera. This intermediate task, moving the camera to follow the insect, is the fundamental difference between the existing system and the primary customer’s proposed system. Every picture taken by the camera will have to be accompanied by the global position of the camera relative to the arena. In this fashion, the global position of the bug will be determined by the vector sum of the camera position and the insect position relative to the camera frame.

Benefit of Solution

The benefit of an insect video-tracking device is very apparent. If an automatic system is devised, the computer will take care of the otherwise tedious and laborious task of moving the camera’s field of view to always encompass the insect. This, of course, assumes that the insect will be roaming in an arena much larger than that of the camera’s field of view. One of the disadvantages to this proposal is that the only one insect can be tracked at a time with the given camera and computer. The customer has already made it clear that this “disadvantage” is acceptable, however.

Method

Our Strategy will be to gain an overall knowledge of the project as awhole and then strive for a solution by researching, benchmarking, and defining the customer’s wants and constraints. Finally, using an iterative design synthesis process, our team will generate the best solution to satisfy our customers.

Mission Statement

Our mission is to design, refine and construct an insect video tracking system to be used with agricultural research, thus providing our customers with a creative, realistic and performance-based solution.

Customer Wants

The customer wants are organized in spreadsheet form, as shown in Appendix A. First, the customers were listed and ranked by their importance to the project. There are six customers, so the most valuable customer was ranked “6”, and the least valuable customer was ranked “1”. Next, the wants of each customer are tabulated in order of importance to complete the table. The most valuable wants to a customer receive a 0.45 weight, while the second most valuable wants to a customer receive a 0.25 weight, the third, fourth, and fifth are shown across the top of the chart in Appendix A. By multiplying the rank by the weight of the customer wants, each want will have an “importance” value. The “importance value” of any wants shared by more than one customer are added together to form a composite want“importance value”. Finally, these “importance values” are sorted, and the wants that receive the highest scores become the most important wants. Appendix B shows the top ten wants arranged in order of importance and their corresponding “rates of importance”.

Constraints

The following constraints were directly imposed by the customer and by New Castle Design Associates. Theseitems are nonnegotiable and must be satisfied.

1. The project must be completed by the end of school year.

This time constraint is a class requirement.

1. The project expenses must remain within several thousand dollars.

The customer has limited our funding.

1. The system must cover an area of onemeter squared.

The customer has made it clear that a one meter squared area must be met in order for the new solution to be more valuable than the existing solution.

1. Two-dimensional tracking system.

The existing equipment uses a two dimensional arena, which is to be used in any new system, as required by the customer.

1. Work area must occupy only Stearns Laboratory.

No equipment may be removed from the customer’s laboratory.

Benchmarking

There is an abundance of material pertaining to real-time electronic imaging, pattern recognition, and motion tracking available. While not all of it has application to the research of insect behavior, much can be learned from the various systems that are being used. Some examples, with references, are listed below.

Existing Video Tracking System

Keith Hopper, USDA & University of Delaware

Topics: Image Processing, Pattern Recognition

Perhaps the most important benchmark is the system that is currently being used. The movements of the insects, which are held within six 4-centimeter diameter arenas, are captured using a fixed CCD camera coupled with a computer. A record of the insect movements is then written to a data file as a time history of two-dimensional space coordinates. A computer program digitizes the CCD analog image in real time. Another program computes the discrete speeds and the average speeds of all six insects simultaneously.

There are, unfortunately, a few disadvantages to this system. The resolution of the camera prohibits the tracking of these insects over much larger areas. Also, it has been found that the 4-centimeter diameter arenas confine the insects too greatly, thus altering their otherwise “natural” behavior.

Vertical Spin Tunnel

NASA Langley Research Center

Topics: Image Processing, Pattern Recognition, Motion Control

In an effort to study the vertical spin and stall characteristics of aircraft, the Langley Research Center has developed a specialized upward flowing wind tunnel in which they place a scale model aircraft. The model is not rigidly attached to any external supports and is allowed to move as physics dictates. Supposedly, this is to simulate the conditions of a vertical spin. An “Automated Video Tracking System” is used to follow the unpredictable motion of the model. The system consists of a zoomable digital camera mounted on a biaxial pivot. The apparatus is controlled via a computer, which adjusts the two angles and the zoom factor.

Apparently, the researchers and Langley are not concerned with locating the object in space, but in viewing it in focus so that it’s orientation with respect to the airflow can be seen. True 3-D spatial location would require two such devices working in tandem, much like human sight.

This type of system is probably too complicated for the needs of this project. Our problem is 2-D tracking only, which requires only one camera, linear motion, and minimal focusing. Nevertheless, significant similarities exist among the two situations: specifically the type of the control system needed and the variables that are involved.

Biorobotic Vision Group

Centre for Visual Sciences

Topics: Image Processing, Pattern Recognition, Motion Control

Interestingly enough, the research at the Centre for Visual Sciences involves both insects and machine vision. They are studying the way that insects see and process visual information in an attempt to model it electronically for the purpose of autonomous navigation. The interest stems from the fact that insects can perform feats of image processing that rival today’s best computers.

This is relevant because basic pattern recognition will be an integral part in any system that tracks the motion of an object. The Centre for Visual Sciences has the advantage that their algorithms are probably simplified yet effective, being that they are modeling insect vision. This might help to reduce overhead processing time; thus increasing the time resolution.

DARPA Image Understanding Program

Texas Instruments Corporate Research Laboratories

Topics: Image Processing, Pattern Recognition

The research at the Image Understanding Program is primarily focused on high-level pattern recognition for use in surveillance or battlefield situations. One of their objectives is to automatically detect, recognize, and track objects regardless of the visibility, vegetation cover, or weather conditions. This has application both in crime prevention and control as well as in military tactics.

Again this is relevant due to the pattern recognition issues. Higher-level object recognition and semantics, however, are too much for our project, which has only one object that is in plain view.

Semi-Automated Film/Video System

Amerinex Applied Imaging

Topics: Image Processing, Pattern Recognition

Amerinex Applied Imaging has developed a software package, currently in use by the US Navy, to track and record the motion of rigid bodies. Fighter landings on aircraft carriers are of particular interest to the Navy, although it could be used for other object as well. There is not much emphasis on the subject of pattern recognition but it mentions that images are recorded and reviewed by a human operator at a later time.

This is of interest because of the raw image processing and data recording capabilities of the software.This is a task, which our final product must perform. After speaking with our main customer, however, it was discovered that a maximum frame rate of about 1 Hz is necessary for the problem. This is not a high demand at all and should be within the capability of most personal computers.

CAD(Computer Aided Design) Plotter Mechanism

Topics: Motion Control

The CAD plotter, of the type used to produce engineering drawings, contains within it the mechanisms necessary for two-dimensional motion control. The plotter pen can move back and forth along its own path while the roller can be rotated to a specific angle. The plotter receives basic commands such as “draw a line from point A to point B” and moves these components accordingly. This type of movement is similar to what may be necessary for the movement of the camera in our project and the concept could easily be carried over.

PC(Personal Computer) Mouse

Topics: Motion Control

ThePC mouse, a common computer peripheraldevice, contains within it mechanisms that relate to two-dimensional motion control. The mouse converts mechanical movement into a set of electronic coordinates. This differs somewhat from the requirements of our project. Our projectmay need toconvert electronic coordinates into a mechanical position forthe camera. While both processes are obviously different, the mechanisms used in both systems are very much alike. Within the PC mouse, a rubber ball rolls against the mouse pad, and two potentiometers. The mouse pad provides enough friction so that when the mouse is moved, the ball rolls in the direction of the mouse. The two potentiometers, at right angles to each other, roll only when the ball is rolling in their direction. By moving the potentiometers, the voltages passing through them changes in such a way that they can be directly related to the position of the mouse with respect to themouse pad.

Goals and Timing

The goals for this project have been organized chronologically using a Gant chart. The Gant chart is shown in Appendix C.

Estimated Budget

The estimated budget for this project is not explicitly defined. However, our primary customer is willing to spend several thousand dollars on the solution. The primary customer will evaluate the cost versus need for each proposed component of our system and provide us with money for those components that seem worthwhile. The primary customer’s existing equipment is also available for our use.

Team Ground Rules

Below is a list of the team ground rules.

1. Be open to constructive criticism.
2. Attend all meetings.
3. Be polite and courteous to all team members.
4. Be prepared for all meetings.
5. Accept mutual responsibility among team members.
6. Be willing to compromise.
7. Keep notebook/log updated at all times.
8. Have fun.
9. No domination allowed.
10. All members must contribute equally.
11. Put customer needs first.
12. Maintain a professional relationship with the customer.
13. Share responsibilities - no task focusing.
14. Follow the code of ethics as directed by class policy.

15. Focus on accomplishing milestones ahead of time.

16. Update the group schedule by the hour.

1. Initiate new tasks - do not procrastinate.

Appendix A

Appendix B

Appendix C

1