Project CompletionProject CompletionECE 496 Fall 2002ECE 496 Fall 2002

GyrobotGyrobotTeam DTeam D

OutlineOutline The Group/Team The Project Design Specifications Design Approach Main Model / Encoders Balance Control Swing-up Control Hardware Challenges Software Challenges Performance Summary Questions

TeamTeam

Ray Price – Team LeaderRay Price – Team Leader David Epting – Hardware David Epting – Hardware

Designer / WebmasterDesigner / Webmaster John AbbottJohn Abbott -- Hardware -- Hardware

Designer / Presentation Designer / Presentation ManagerManager

Matt Vaughn – Lead Matt Vaughn – Lead Software Designer / Software Designer / Electronics TechnicianElectronics Technician

Cyrus Griffin – Software Cyrus Griffin – Software Designer / PhotographerDesigner / Photographer

The ProjectThe Project The Gyrobot is an underexcited The Gyrobot is an underexcited

pendulum, consisting of a pendulum, consisting of a single link (arm) with a flywheel single link (arm) with a flywheel driven by a dc motor mounted driven by a dc motor mounted at the free endat the free end

The Gyrobot was to include a The Gyrobot was to include a control algorithm that uses the control algorithm that uses the generated inertia of the generated inertia of the flywheel to cause the pendulum flywheel to cause the pendulum to invert and balance with the to invert and balance with the flywheel at the 12 o’clock flywheel at the 12 o’clock PositionPosition

Project SpecificationsProject Specifications The Gyrobot had to fit the following criteria:The Gyrobot had to fit the following criteria:

Must comply to the mechanical specification of Must comply to the mechanical specification of thesis by Adrian Jenkyn Lee out of the thesis by Adrian Jenkyn Lee out of the University of Illinois at Urbana-Champaign.University of Illinois at Urbana-Champaign.

Must utilize motor/flywheel inertia to invert Must utilize motor/flywheel inertia to invert pendulum and then balance.pendulum and then balance.

Utilizes Simulink RTW controller.Utilizes Simulink RTW controller.

Design ApproachDesign Approach

Specified / developed hardwareSpecified / developed hardware Procured hardwareProcured hardware Assembled GyrobotAssembled Gyrobot Tested interfaces (encoders and analog Tested interfaces (encoders and analog

output)output) Compiled Encoder / Main RoutineCompiled Encoder / Main Routine Compiled Balance Routine and TestedCompiled Balance Routine and Tested Compiled Swing-up Routine and TestedCompiled Swing-up Routine and Tested

Design ApproachDesign Approach

Gantt ChartGantt Chart

Main Software Model

The main software model combined the encoders, swing-up and balancing algorithms.

Position EncoderPosition Encoder

Used to produce arm position (theta\1 Used to produce arm position (theta\1 from 0 to 2pi and a theta1 from –pi to from 0 to 2pi and a theta1 from –pi to

pi) also produced the arm velocity pi) also produced the arm velocity theta1dot.theta1dot.

Motor EncoderMotor Encoder

Converts number of swings to radian and filters to produce a theta2dot—the flywheel velocity.

SoftwareSoftware Pd Balance Control - The ModelPd Balance Control - The Model

Balance ControlBalance Control

Important Variables:Important Variables: Arm position Arm position

(theta 1)(theta 1) Arm velocity Arm velocity

(theta1dot)(theta1dot) Flywheel velocity Flywheel velocity

(theta2dot)(theta2dot)

Balance ControlBalance Control

Arm PositionArm Position Added enough energy to move mass of Added enough energy to move mass of

assembly to the highest position--Fighting assembly to the highest position--Fighting gravitygravity

Gain of kp = 3.375 was used based on center Gain of kp = 3.375 was used based on center of gravity and mass of the mobile assembly of gravity and mass of the mobile assembly (motor, flywheel, arm, shaft)(motor, flywheel, arm, shaft)

Balance ControlBalance Control

Arm VelocityArm Velocity As the arm approached vertical it should slow. As the arm approached vertical it should slow.

Arm must be able to fight acceleration if it falls Arm must be able to fight acceleration if it falls

away from vertical.away from vertical. Gain of kd = .72 based on rotational inertia of Gain of kd = .72 based on rotational inertia of

the whole mobile system. the whole mobile system.

Balance ControlBalance Control

Flywheel VelocityFlywheel Velocity The flywheel stopped when the arm is The flywheel stopped when the arm is

balancing.balancing.

Gain of k = .0006 was small, in effect Gain of k = .0006 was small, in effect creating an under-damped system.creating an under-damped system.

Gain is negative 175 (after the summer) to Gain is negative 175 (after the summer) to bring the speed of the flywheel to zero bring the speed of the flywheel to zero (instead of slowly ramping up).(instead of slowly ramping up).

Swing Up ControlSwing Up Control

Sinusoidal model from thesis was used Sinusoidal model from thesis was used because:because:

SmootherSmoother Faster due to harmonicsFaster due to harmonics Less bouncing in controls compared to other Less bouncing in controls compared to other

proposed methodsproposed methods

Swing Up ControlSwing Up Control

Sends a sinusoidal signal to the motorSends a sinusoidal signal to the motor

Motor switches polarity via a switch when the Motor switches polarity via a switch when the arm reaches zero velocityarm reaches zero velocity

Theoretically the control effort is supposed to Theoretically the control effort is supposed to slow down as balance is approached, but since slow down as balance is approached, but since we saturated the effort this doesn’t really happen we saturated the effort this doesn’t really happen

Mechanical Design ChallengesMechanical Design Challenges

FlywheelFlywheel ProblemsProblems

Flywheel was not properly centered during milling Flywheel was not properly centered during milling processprocess

Wheel would wobble and eventually flew offWheel would wobble and eventually flew off

SolutionSolution A glue was applied along with the set screwA glue was applied along with the set screw

The glue absorbed most of the vibration, drastically The glue absorbed most of the vibration, drastically reducing the wobblereducing the wobble

Mechanical Design ChallengesMechanical Design Challenges

Pittman MotorPittman Motor ProblemsProblems

While pressing the flywheel onto the motor the encoder was While pressing the flywheel onto the motor the encoder was pushed off-centerpushed off-center

Heat generated during use led to inconsistent performanceHeat generated during use led to inconsistent performance

SolutionSolution A new motor was ordered in exchange for the damaged one A new motor was ordered in exchange for the damaged one

and the flywheel was attached by a set screw instead of and the flywheel was attached by a set screw instead of being pressed onbeing pressed on

Followed a set timing scheduleFollowed a set timing schedule

Mechanical Design ChallengesMechanical Design Challenges

BearingsBearings

ProblemsProblems Bearings were too stiff, generating un-needed Bearings were too stiff, generating un-needed

friction friction

SolutionSolution Bearings eventually loosened up after useBearings eventually loosened up after use

Control Challenges - BalanceControl Challenges - Balance

Problem: Parameter OptimizationProblem: Parameter Optimization Solution: Optimize only 1 control variable Solution: Optimize only 1 control variable

at a timeat a time

Control Challenges - BalanceControl Challenges - Balance

Problem: Limited Pull-up abilityProblem: Limited Pull-up ability Solution: Create a window outside of Solution: Create a window outside of

which the routine does not engage. which the routine does not engage.

Control Challenges – Swing-upControl Challenges – Swing-up

Problem: Recovery time between runs.Problem: Recovery time between runs. Solution: Use a 4-swing swing-up. Solution: Use a 4-swing swing-up.

Allow cooling time.Allow cooling time.

Control Challenges - SwingupControl Challenges - Swingup

Problem: Inconsistent effort window.Problem: Inconsistent effort window. Solution: Set window each day based on Solution: Set window each day based on

the temperature of the room, and the the temperature of the room, and the temperament of the gyrobot.temperament of the gyrobot.

Control Challenges - Control Challenges - TransitionTransition

After a repeatable swing-up was established, there After a repeatable swing-up was established, there were still problems with the transition to balance. were still problems with the transition to balance.

Problem: Inconsistent room temperature.Problem: Inconsistent room temperature. Solution: Practice the routine enough times on Solution: Practice the routine enough times on

competition day to get a “feel” for cooling time.competition day to get a “feel” for cooling time.

PerformancePerformance

Final CompetitionFinal Competition Fastest time – 2.47 seconds (2Fastest time – 2.47 seconds (2ndnd Place) Place)

Bonus DayBonus Day 9 out of 109 out of 10

SummarySummary

Were able to build a gyrobot device and Were able to build a gyrobot device and the associated control structure that would the associated control structure that would invert and balance the gyrobot penduluminvert and balance the gyrobot pendulum

Able to balance and resist impulsive forces Able to balance and resist impulsive forces against the deviceagainst the device

Able to “swing-up” in 4 swings.Able to “swing-up” in 4 swings.

Questions?Questions?