Project “George” Design Proposal - Concordia University



Project “George” Design Report

EGR 345/101

December 4th, 2004

Presented to: Drs. Jack & Farris

Presented by: Team #5

Justin Bjorum

Richard Jansen

Mary Kundrat

Rick Lopez

Jeff Schober

Nick Hathaway

Executive Summary

This document discusses the main concepts and design considerations of a two-wheeled self-balancing cart similar to a Segway human transporter capable of following simple to complex paths. Design specifications required that the wheels be less than 152.4 mm in diameter, the handlebars be between 177.8 and 228.6 mm in length, and the entire system be self-contained. The cart is composed of lightweight aluminum and polycarbonate, which provide a rigid base while maintaining minimum weight. Lightweight right angle 5V DC motors are used to drive the gears, which provide power to the wheels, and at the same time turn the encoders. A C program was written for the Atmega32 microcontroller board to give the robot the ability to self-balance, the ability to implement feedback control, and the ability to steer. The cart uses a gyroscope to measure tilt and keep the payload balanced, while the encoders taken the position of each motor and monitor velocity. Two photo resistors are used as optical sensors to sense the desired path and adjust the cart if it ever strays from this desired path. The estimated cost for the cart was $112.00, and the total weight of the cart is 0.5 kg.

Table of Contents

Project Overview 2

Project Objective 2

Project Description 2

Project Constraints 2

Conceptual Design Description 2

Mechanical Design Aspects 2

Major Components 2

Wheel size/Axle Positioning 2

Power Transfer 2

Electrical Components 2

Circuits and wiring 2

Software Design Aspects and Components 2

System Architecture 2

System Block Diagram 2

Calculations and Differential Equations 2

Free Body Diagrams 2

Stress Calculations 2

Bracket 2

Wheel 2

Final Project Budget, Bill of Materials, Weight Inventory, & Performance Prediction Calculation 2

Changes to the Initial Design to compensate for Problems Incurred 2

Scilab Simulation 2

C Program 2

Fabrication Process 2

Testing Procedures 2

Conclusions 2

Recommendations for Improvement 2

Project Overview

Project Objective

The objective of this project is to produce a complex engineering system from concept to completion, by utilizing proper project management skills, teamwork, and engineering principals.

Project Description

“George” is designed to be a working prototype of a Segway Human Transporter. The Segway HT is a two wheeled transportation device that allows the rider to stand erect while riding. Each wheel has its own motor, and they, together, propel the device and balance the rider. The prototype itself must be self contained, including the power supply, and must be functionally similar to the Segway, excluding the user controls and displays. The prototype must also incorporate line following capabilities and higher speed motion.

Project Constraints

Since the project focus is on developing a Segway prototype, many constraints have been defined in the project description. The following list describes all of the constraints that were taken into consideration in determining the final conceptual design:

1. Device has to be ≤ 6 inches tall

2. The wheel diameters have to be between 0.5 and 6.0 inches tall.

3. Robot must be able to follow both simple and complex paths.

4. The robot must be self-contained, including its power source.

5. The robot must be able to slow down and stop normally.

6. The robot must contain a scaled version of the Segway’s foot plate that can support a water mass weighing up to 2kg.

7. The robot must include a vertical post that is 7 to 9 inches tall.

8. The robot itself must be free to rock on its wheels over a range of ± 60° without falling.

9. The total part and equipment cost must be ≤ $200.00

10. The total mass of the system, without the additional water mass must be ≤ 1kg.

11. The model robot must be built using a clearly justified and supported design process, which will be judged in a competition.

The cart must also be operated by a simple on/off switch. Once the cart is turned on, it must be able to move forward at full speed when it is traveling directly over a line, but must slow down and stop when not directly over a line. The line will be composed of black electrical tape, which can be easily detected by a form of optical sensor. The cart itself must be able to navigate over both simple and complex paths, involving straight lines, corners, and curves. In order to verify that these constraints are met throughout the duration of the project, the following ‘score’ equation, shown in Figure 1, has been included in the project description and will be calculated each time a factor in the equation changes. At the completion of the project, all of the projected scores will be compared for accuracy.

[pic]

Figure 1: Performance Prediction Equation

Conceptual Design Description

Mechanical Design Aspects

Major Components

The major components involved in the prototype design include: two 5V miniature gear motors, two lightweight aluminum plates, two PVC wheels, two polycarbonate (PC) brackets, a gyroscope, two photo resistors, a CTS 16 position non-detent mechanical encoder, two nylon spur gears, an ATMega32 microcontroller board, nylon bolts, a power switch, four steel bearings, wires, and a hollow aluminum shaft. The details pertaining to each component, such as cost, weight, purchasing status, and positive and negative aspects, are displayed in Appendix A.

Wheel size/Axle Positioning

A low center of mass will allow for a more stable system, therefore axle positioning will be relatively high above the base plate. This, along with the gyroscope, will contribute to the system’s ability to balance. Positioning is mainly determined by the size of the wheel and the width of the chassis base plate. The angle of the chassis base plate must be permitted to tilt. For 12.7mm of clearance and the required tilt angle of Ө>60°, it was determined that 80mm would be the optimum width for the base plate. To accommodate the high axle positioning, twin pillars were designed to bolt to the base plate. These pillars will position the axle approximately 90.5mm above the base plate, allowing for the 12.52mm of ground clearance. The pillars are designed to be as light weight as possible and have a second tier support, (facing inwards) which will be mounted to the top plate. Figure 2 shows an isometric view of the assembly.

[pic]

Figure 2: Isometric View of "George"

An isometric view of one of the wheels is shown in Figure 3. The wheels are made out of PVC, and unnecessary material was trimmed from the insides of each wheel, leaving three webs, which are necessary to support the weight of both the cart and the payload.

[pic]

Figure 3: Isometric View of the Wheel

The isometric view of one of the brackets is shown in Figure 4. The two brackets combined support the top plate as well as the 2kg mass being applied.

[pic]

Figure 4: A Side Bracket

An isometric view of the top plate, which will hold the payload, is shown in Figure 5.

[pic]

Figure 5: The Top Plate

CAD drawings of the assembly, as well as all of the individual parts are included in Appendix B.

Power Transfer

Adequate torque was provided to each wheel to allow the 0.5 kg cart and its 2kg payload to accelerate and decelerate smoothly. The design utilizes two gears to transfer the torque from each motor to its axle. Each moor shaft has a gear pressed onto it, which will in turn transfer the torque to a similar gear pressed onto the axle. No gear differential is needed due to the slow angular velocity of each wheel with respect to its motor. A lower center of mass will also be achieved as a result of positioning the motors on the base plate. All unnecessary materials in the base plate have been removed to eliminate unwanted mass.

Electrical Components

Circuits and wiring

For a successful balancing prototype, it was determined that the design would consist of two encoders that will read the position of both the left and right motors. The voltage being returned from two separate photo eyes, one for each motor, will be used to determine whether the cart is on the desired path. One gyroscope is used to monitor the yaw rate of the aluminum plate on which the 2kg mass will be placed. The measured velocity is then integrated to determine the vertical position of the plate. This helps feed the actual position of the plate back into the controller board to support the self balancing efforts of the mechanism. The L293 is an integrated circuit motor driver that is used for simultaneous, bi-directional control of two small motors. A picture of the L293D is shown in Figure 6.

[pic]

Figure 6: The L293D Motor Driver

A complete circuit schematic graphically portraying all of the circuit elements is shown in Figure 7.

[pic]

Figure 6: Complete Circuit Schematic with R = 10Ω

Software Design Aspects and Components

System Architecture

The block diagram for “George” is shown in Figure 8. It shows how the carts actual position is fed back into the Atmega32 controller board and used, along with deadband compensation and PWM to input the desired position to each motor.

[pic]

Figure 7: System Architecture

System Block Diagram

The system’s block diagram describes the order in which things will happen. It is shown in Figure 9.

[pic]

Figure 8: System Block Diagram

The control system takes the actual angular position of the cart, along with the position of the top plate and payload that is found from the gyroscope, and implements the A/D converter to calculate the output command for the position control. The difference between the position error and commanded wheel speed is then feed into the proportional gain controller to yield the proper output command for the position control. The difference between the command for position control and the command for angle control is then fed through the deadband compensation function to correct for whatever deadband that might exist. This output is sent through the PWM functions, and the resulting effective motor control voltage is sent through the motor producing the commanded wheel speed.

Calculations and Differential Equations

Free Body Diagrams

To properly simulate what would happen dynamically to the payload, free body diagrams were made for the 2kg mass, and the summation of forces produced equations for both translation and rotation of the mass. The free body diagrams for the 2kg mass are shown in Figure 10.

[pic]

Figure 9: FBDs of the 2kg Mass

The calculations that support these diagrams that were used to develop the system’s state equations are shown in Figure 11.

Calculations:

[pic] = Force from the motor

[pic] = Force of axle

[pic] = Mass of cart

[pic] = Mass of wheel

[pic] = Angle of cart displacement

[pic] = Angle of wheel displacement

[pic] = Translational displacement

[pic] = Radius of the wheel

[pic]= Resistance of the motor

[pic]= Distance from C.O.M to center of rotation (on cart)

Reference:

A.) [pic]

B.) [pic]

C.) [pic]

D.) [pic]

E.) [pic]

CART:

Reference equation C

[pic]

[pic]

[pic] (1)

[pic]

[pic] [pic]

[pic]

[pic]

[pic] (2)

WHEEL:

Reference equations A and E:

[pic]

[pic]

[pic]

[pic]

[pic] (3)

Get equation 1 in terms of [pic]:

Reference equation D:

[pic]

[pic]

[pic] (4)

Equation 4 into equation 3:

[pic]

[pic] (5)

Insert motor parameters:

[pic]

[pic]

[pic] (6)

Equation 6 into equation 5:

[pic]

[pic]

[pic] (7)

State Equations:

[pic]

[pic]

[pic]

[pic]

Figure 10: Calculations and State Equations of Motion for the FBDs

Stress Calculations

After analyzing the payload, it was important to verify that the stresses created by the 2kg mass did not exceed the maximum allowable stress of all of our parts and materials. The stress, failure, and factor of safety calculations are shown in Figure 12.

Bracket

The side brackets of the chassis are used to connect the chassis to the wheels. The cart is symmetrical and the C.O.M of the cart and applied load is assumed to be in the center of the cart chassis. Therefore, each side bracket must support half of the 2.2 lb cart plus half of the 4.4 lbs of the applied load.

Side Bracket Hole:

Shear Stress

[pic]

Where,

P = Force being applied ( 6.6 lbs / 2 )

A = Cross sectional area where the hole is located

[pic] where,

c= Distance across section – hole diameter (inches)

t= Thickness of bracket (inches)

K = 2.25 = Stress concentration factor

K was referenced from ( Mechanics of Materials, James M Gere, pg 142.)

Therefore the equivalent shear stress can be calculated as:

[pic]

[pic]

This stress is not excessive for the plastic brackets. The yield strength for Plexiglas is between 7.1ksi and 10.6ksi. The factor of safety ‘n’ can now be calculated as follows;

[pic] where,

[pic]= yield stress

[pic]=shear stress

[pic]

By using the lowest possible yield stress the minimum factor of safety was found to be 98.5. The factor of safety gives assurance that the bracket will not fail under the given conditions.

Bearing Stress

The bearing stress for the bracket will be calculated using the same force applied as in the shear stress calculation. The equation for bearing stress is as follows:

[pic]

Where,

P= applied force (3.3 lbs)

t = thickness of part (.25 inches)

D = diameter of the hole (.5 inches)

[pic]psi

The factor of safety can be calculated using the same yield stress.

[pic]

The factor of safety indicates no reason to believe this part would fail under the given conditions.

Wheel

The wheels of the cart are a vital design consideration. The wheels support all of the weight of the cart and applied load. A stress and deformation calculation was done in order to assure that the wheels would function properly as designed.

The first stress calculation will be on the axel support of the wheel.

Shear Stress

[pic]

Where,

P = Force being applied ( 6.6 lbs / 2 )

A = Cross sectional area where the hole is located

[pic] where,

c= Distance across section – hole diameter (inches)

t= Thickness of bracket (inches)

K = 2.5 = Stress concentration factor

K was referenced from ( Mechanics of Materials, James M Gere, pg 142.)

Therefore the equivalent shear stress can be calculated as:

[pic]

[pic]

This stress is not excessive for the PVC wheel centers. The yield strength for Rigid PVC is between 4.2ksi and 5.8ksi. The factor of safety ‘n’ can now be calculated as follows;

[pic]

By using the lowest possible yield stress the minimum factor of safety was found to be 63.6. The factor of safety gives assurance that the wheel will not fail under the given conditions.

Tensile Stress

The tensile stress will be calculated for the webbing that connects the wheel hub to the outer ring of the wheel.

The equation for the max stress will be calculated from the following equation:

[pic]

[pic]= 52.8 psi

This stress is not excessive for the webbing on the wheels. The factor of safety ‘n’ can now be calculated as follows;

[pic]

Elongation / Deformation

The deformation of the wheel should be calculated because PVC is a ductile material and can be subjected to plastic deformation or necking in regions of high stress. If deformation occurs it would cause error in many calculations and cause the cart not to perform as designed. The webbing was determined to be the area of most concern because it has a long thin section that would most likely be deformed. The deformation will be calculated by the following equation:

[pic]

where,

L = length of axial loaded member (inches)

E = Modulus of elasticity (psi)

A= Cross sectional area (inches)

P= Force applied (lbs)

Given the dimensions and the material properties of PVC the deformation was found to be.

[pic]

The deformation of 0.00041 inches is adequate given the design criteria and should not hinder performance or functionality of the cart.

Figure 11: Stress Calculations

Final Project Budget, Bill of Materials, Weight Inventory, & Performance Prediction Calculation

The project budget, along with an updated weight inventory, bill of materials and performance prediction calculation are shown in an Excel spreadsheet in Figure 13. This made it easier for the group to change aspects of the budget as the project progressed.

[pic]

Figure 12: Project Budget, Weight Inventory, & Performance Prediction Calculation

Excluding the 2kg mass, the overall estimated weight of “George” is 0.50kg and the total cost is approximately $120. The weight inventory has been frequently updated throughout the project to ensure the total weight of the system remains less than or equal to 1kg. Since the overall fabrication and design of “George” produced a quality cart, fairly high build and theory qualities were used to compute the performance prediction calculation, despite that these values will ultimately be determined by the judges.

Changes to the Initial Design to compensate for Problems Incurred

During the fabrication and building process, many problems were incurred making changes to the initial design necessary. Many of these complications were dealt with quickly, not delaying the progress of the project. There were changes that took close to a week to completely implement. These high cost (in terms of time) changes carried a ‘trickle down’ effect, where not having the correct part prevented the testing of other parts which in turn was also problematic. Almost every part on the robot was modified and additional parts were created, simply to compensate for specific problems.

Before fabrication, the original Lower Chassis Plate (LCP) was modeled using ProEngineer design software. After stress calculations were completed for the solid plate of aluminum, it was decided that the LCP could have a great deal of material cleared out without compromising the performance of the chassis as a whole. A large clearing was implemented underneath the AtMega32, since no parts needed to rest there. The batteries were originally positioned on the LCP, however the creation of two locator holes in the plate allowed for the batteries to be recessed. This helped the robot in more than one way, by not only reducing the overall weight, but also by lowering the overall center of mass.

The same method of material removal was implemented in the design of the Upper Chassis Plate (UCP). Stress calculations showed that an increased factor of safety could be achieved by adding an additional cross member (bringing the total to two). This change was again made in a virtual environment (ProE) and required no extra fabrication time.

The final major components of the chassis are the pillars, which also needed additional features for it to be beneficial to our design. A central ‘spine’ was added within the ProE assembly to accommodate a motor shaft support bearing. This bearing prevents any additional perpendicular axis torque from being transmitted to the gears by the wheels. The Pillars were CNC-milled a total of five times, due to machine control errors. After three erroneous parts it was decided that extremely slow feed rates on the Micro-Kinetics™ machines were necessary to maintain a proper radius. This waste of time cost forward progress of the project to be pushed back by an entire day.

The gearing of the robot was problematic from the start; a motor relocation was implemented after the original gears arrived via USPS (United States Postal Service, 5-day shipping time). Obviously the gears were scrap before they were taken out of the box, luckily another group was in need of the same gears and able to purchase them. A second set of gears was ordered with a higher pitch, however the color scheme of the robot would suffer. The second set of gears turned out to have completely random coloring, two red and two green, neither of which would coordinate with a blue and silver robot. To compensate, three different brands of spray paint were tried, none of which adhered to the plastic. On top of the color problem a bigger problem arose. The inner diameter of the gears was over-bored on the drill press and not correctly centered. This led to binding issues and overall elimination of the gears. Finally a third set was ordered, in which gears with a lower pitch diameter were selected to prevent any chance of binding. The company (ProBelay Gears) offered two different sizes of small gears which would work on the robot, a 10 tooth and a 12 tooth, so to be extra careful, a set of each was ordered (along with the larger 24 tooth gears). The robot was equipped with the 12 and 24 tooth gears, conveniently molded with blue nylon. This setback cost roughly a week in fabrication time.

On the electrical side of the robot, the gyro chip became a problem. Initially two gyros were donated by Analog Devices ™ which use iMEMS (integrated Micro Electro Mechanical System) technology to measure rate of angular displacement. This chip has proved to work well, however there were problems with the original wiring. After connecting AVCC and AGND to a 5V DC power source and the RATEOUT and AGND to a DMM, no angular rate was being produced, instead a simple voltage divider was returned. After consulting the operation manual for the chip once again, it states:

“Note that the analog supply voltage and charge pump supply voltage (AVCC and PDD) are not connected on the ADXRS150EB and that users must connect these as appropriate to their application” (Analog Devices, 1).

The problem with this statement was that it did not specify what type of application would validate this, so through trial it was found to be the problem. The pins were connected and the gyro produces consistent and accurate angular rate results.

Aside from these fairly detailed changes, there were a few small items which needed either positioning brackets or a simple modification. These included:

• Wheels – added a set screw to eliminate possible axle slip

• Photo-Resistors – Locating brackets had to be fabricated after wiring the

AtMega board.

• On/Off Switch – Wire breakage issues on the battery terminal made it necessary to fabricate a small bracket to locate the switch off of the UCP.

Programming the Controller

Scilab Simulation

The purpose of the Scilab simulation is to display what the motor should do based on the motor parameters and state equations of the system. The plots produced by Scilab represent the movement of the payload as well as the relative velocity of the cart itself. The Scilab program is in Appendix O.

C Program

The C program for the control of the system implements the system’s block diagram shown in Figure 9. It is expected that the C program will use that block diagram to control its input to the motor and thus control the speed and position of the cart. The final program takes the output from both photo eyes and adjusts the desired speeds of each motor. Then, the output from the encoders is used to adjust the actual position and velocity of each motor. The desired and actual speeds are compared in the controller function of the feedback loop, which subtracts the two values, multiplies their difference by proportional and integral gains, and sends the resulting value to the deadband compensation function. The deadband compensation allows the cart to overcome both static and kinetic friction, decreasing the amount of voltage needed for the cart to start moving. Without utilizing deadband compensation, the cart would not initially accelerate when we wanted. Instead, more voltage was supplied and the result was motor noise and no other motor activity until the voltage supplied was powerful enough to overcome either type of friction. Once the deadband compensation limits were determined, and the function was enabled, the motor was able to overcome the friction instantaneously, and less motor noise was produced. After the deadband compensation program, the adjusted values for each motor are sent through the “v_output” function. The purpose of the v_output function is to adjust the direction of the rotation of the cart depending on whether the cart’s speed is positive or negative. If a negative value is sent to this function, the adjusted voltage becomes negative and the cart travels in the reverse direction. To test this function, a positive value was sent to the v_output function for the left motor and a negative value was sent to the v_output function for the right motor. The result was the left motor traveling forward (cw) and the right motor traveling backward (ccw). This activity verified that the v_output functions worked properly. Once through this function, the adjusted values are sent through the Pulse Width Modulation functions and sent to each motor. The entire feedback loop is located in the IO_update function, so that the feedback from the motors is monitored on a more frequent basis and the cart will react much faster. The C program is shown in Appendix M.

Fabrication Process

Strategies for building and designing this complex system are listed below in bullet format. These are the machining and assembly techniques that we plan on using for fabrication.

a. Machining

1. The base plate and the 2nd tier are CNC-milled aircraft-grade aluminum.

2. The pillars and wheels are CNC-milled light-weight plastic.

3. The axles and bushings are aluminum and Delrin (UHMW), respectively. Both parts were machined on a lathe.

b. Assembly

4. All mounting holes on the base plate were tapped.

5. The 2nd tier supports were drilled and tapped.

6. The Delrin bushings must were pressed into the pillars.

7. Each pillar was fastened to the base plate via two plastic bolts.

8. Each aluminum axle was pressed into a wheel.

9. The axle was then inserted into the bushing (slip fit) and secured by pressing on the larger diameter gear.

10. Each motor was loosely mounted onto the base plate and the gear positioning was marked on the shaft to ensure proper alignment once the gear was pressed on.

11. Pressed small gears on motors taking special care they were aligned properly.

12. The motors were securely fastened to the base plate. Care was taken to ensure proper contact with gear teeth, and motor height was shimmed when necessary with feeler gauges.

13. The wheel speeds were tested to be sure the system was assembled correctly at that point in time.

NOTE:

No two motors are the same; there was a certain amount of variation no matter how precisely the cart was assembled. This test was simply to verify there was no binding in the gears or added friction due to the bushings. The wheel speeds were close, but not exactly the same. These differences were taken into consideration with the software.

Each motor was then hooked up to a power source and supplied exactly 10 volts of power (a DMM must be used to verify this). The motors’ speeds were then measured with a strobe tachometer. There were no major differences in rotational velocity, meaning that assembly errors were minimized during fabrication. Both wheels were exhibiting similar rotational characteristics, meaning the fabrication process could continue.

14. The AtMega32 microcontroller board was then bolted onto the base plate using four plastic bolts and spacer cylinders.

15. Photo-voltaic sensors were placed through the pre-drilled holes in the base plate, taking special care to isolate leads from aluminum base.

16. The accelerometer will be fastened with hot glue, in the vertical position, to pre-measured point on base-plate.

17. Two 5V batteries were wired to the L293D motor driver. The batteries were not mounted in position until the fabrication process was almost complete. This helped with final balancing adjustments. These batteries were by-passed for testing purposes, but still had to be part of the system to accurately depict behavior.

18. The gyroscope, photo-voltaic cells and encoders (not mounted yet) were wired to AtMega32. All excess wire was then trimmed for weight and aesthetics.

19. The 2nd Tier was fastened to the supports using four plastic bolts.

20. The encoders were then pressed into the inside of the axles, (taking special care not to damage them,) and mounted onto 2nd tier.

21. The batteries were then positioned and fastened onto the base plate in such a way that the cart self-balanced.

22. Faux handlebars were fabricated, then placed into pre-drilled holes on the 2nd tier.

23. Finally the software was uploaded.

Testing Procedures

Once the conceptual design was finalized and fabrication of the different parts was complete, it was necessary to test the assembly to determine how each component would function. The testing occurred in two phases. It was first necessary to test specific hardware components to determine their functions, as well as how their output could benefit the performance of “George.” The second phase involved program testing. Each function in the C program was written to complete a specific task. It was important to ensure that the most important functions, such as the Pulse Width Modulation function and the deadband compensation function, were performing their expected tasks. These functions, along with the adjusted voltage function (v_output) are the foundation for the C program and it is pertinent that they operate properly.

The hardware components of “George” that were tested include the encoders, the gyroscope, and the photoeyes. During the hardware testing phase the pin configurations and the voltages being output to specific pins were also tested to verify that the pin configuration code in the C program was correct. This does not classify as a hardware test, however it was necessary to have the hardware properly assembled, with the other components working to verify that the pin configurations were correct. This was the first test that was performed. The C program was downloaded to the ATMega32 board, and a Digital MultiMeter (DMM) was used to verify the voltages coming out of the pins that were meant to supply other hardware components with 5V logic supply power.

The encoders were the next set of hardware components that were tested. Encoders are meant to monitor position, but in “George’s” case, they are being used to monitor velocity. After each encoder was installed, small functions were added to the C program to allow the user to display the hex values related to each encoder’s position on the monitor. Theoretically, if the encoder rotates clockwise, its corresponding hex value should increase, but if the encoder rotates counterclockwise, its hex value should decrease. To test this, the shaft of each encoder was rotated manually. While each shaft was being rotated, the encoder’s current hex value, or position, was displayed on the monitor. When the encoders were rotated clockwise the values increased and how quickly they increased was dependent on how quickly the shaft was being turned. However, if the shaft was rotated in the other direction, the hex values began to decrease. This helped verify that the encoders were functioning properly and could be used for this application.

The next piece of hardware tested was the Gyroscope. The purpose of the gyroscope is to measure the angular velocity of the cart to the normal axis (the Yaw rate) and use the voltage value proportional to that velocity to adjust the position of the cart, allowing “George” and its payload to balance itself. The gyroscope was first tested independently of the entire assembled system. Positive and negative connections were made and the voltage coming out of the gyroscope (RATEOUT) was measured and monitored using a DMM. As the gyroscope was rotated slowly, the change in voltage was continuous and steady. Keeping the gyroscope level allowed us to determine the system’s steady state voltage (as read by the gyroscope), or the voltage being returned from the gyroscope when the cart itself is completely level. This voltage once again correlates to the system’s yaw rate, which needs to be zero for the system to be completely balanced. The steady state voltage of the system is 2.03 V. When rapidly rotating the gyroscope in either direction, the maximum change in voltage seen by the gyroscope was approximately 4.8V. After determining the range of voltages that would be returned by the gyroscope, this range was used, along with the system’s steady state voltage to develop some computer functions that allow the cart to balance. If the voltage output from the gyroscope was anything other than the steady state voltage, the velocities of each of the motors was adjusted to accommodate the imbalance of the cart, and thus allow the cart to readjust and become balanced. To test this theory, the gyroscope was attached to a bracket that was permanently fixed to the cart. As the gyroscope would constantly read and return its voltage values (which are proportional to yaw rate) motors speeds were increased and decreased to accommodate for the different between the actual voltage seen and the steady state voltage. The closer the actual voltage was to approaching the steady state voltage, the less the system swayed, and the more the system became balanced.

The other important hardware components tested were the two photo resistors, or photoeyes. Each photo eye was mounted on a bracket that is situated on the bottom of the cart. The purpose of the photo eyes is to take the amount of voltage they output (which is proportional to the amount of light they see) and use that voltage to determine whether or not the cart can see the path of black electrical tape. A function was added to the C program that allowed us to display the hex values of each photo eye on the monitor. Viewing these values with the cart situated on different types of surfaces allowed us to determine the relationship between the amount of light each photo eye sees, the resulting voltage value produced by each photo eye, and the hex value of each photo eye which corresponds to the voltage. First, the two photo eyes (already fixed to the cart) were placed over a white sheet of paper. The voltage at that point for each photo eye was recorded, using a DMM, and the hex value of each photo eye was displayed on the monitor. The same process was repeated for when the cart was situated over a tile floor, and for when the photo eyes were situated directly over black electrical tape. It was determined from these tests that when the photo eyes returned high voltages, their hex values were also high and when the photo eyes returned a low voltage, their resulting hex values were lower. It was also determined that the higher voltage values were returned when the photo eyes saw the most light and the lower voltage values were returned when the photo eyes saw the least amount of light. Therefore, when the photo eyes produced low voltages, they were situated on or near the electrical tape, and when the photo eyes produced high voltages, they were not situated near any tape. We used these tests results to develop functions that would ultimately allow us to steer the cart. Since the objective was to have the cart follow a path of electrical tape, our goal was to keep the voltage values that the photo eyes see low, meaning that it always stays on the desired path. In order to do this, we needed the hex values of the photo eyes to also be low. If the hex values increased that means that the photo eyes were seeing too much light and steering away from the desired path. Some if/else statements were created to control and adjust the hex values of the photoeyes depending on the values the photo eyes were returning, to keep the cart on the desired path.

Conclusions

Upon completion of the Segway contest and judging, conclusions will be made on the overall project performance, specific test results, and group and individual evaluations.

Recommendations for Improvement

After reviewing the performance of the cart some improvements could be made. The wheels of the cart could be made stronger by removing less material out of the centers and thickening the webbing to the axle. By adding material to the wheels, they would be more rigid and be less susceptible to deflection. This would help reduce the possibility of the bottom of the cart hitting or rubbing against the ground. Another improvement that cold be made would involve relocating the serial port of the circuit board so that the board does not have to swivel out when making the serial port connection. This could be done by splicing wirers to the existing serial port and running them to a more appropriate place on the board, and then connect a serial port connection to the end. Another possible improvement involves increasing the ground clearance of the cart to ensure its mobility on several different kinds of terrains. This would allow the cart to travel much easier on carpet or any other bumpy terrain. It would be beneficial to reducing the height of the side pillars in order to move the bottom plate up and increase the ground clearance. Finally, the wiring placement is satisfactory, but could be improved to provide better aesthetics and clarity in the wire functions, definitions, and configurations.

Appendices

Appendix A: “George” Base Plate

Appendix B: “George” Assembly Unexploded

Appendix C: “George” Assembly Exploded

Appendix D: “George” Side Bracket

Appendix E: “George” Top Plate

Appendix F: “George” Wheel

Appendix G: CAD Drawing of “George” Exploded

Appendix H: Chassis Assembly Drawing Unexploded

Appendix I: CAD Drawing of Base Plate

Appendix J: CAD Drawing of Side Bracket

Appendix K: CAD Drawing of the Top Plate

Appendix L: CAD Drawing of a Wheel

Appendix M: C Program

Appendix O: Scilab Simulation Program and Plot

Appendix P: Updated Gantt Chart

Appendix Q: Receipts

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download