-Pujan Shrestha Class of 2015

Smartphone Inertial

Navigation

INTRODUCTION• Inertial navigation system (INS) use accelerometers, gyroscopes to

measure the position, orientation and velocity of a moving body.

• The calculation of the position and velocity is entirely Internal.

• Conventional accelerometers and gyroscopes have been increasingly replaced by microchips

WHAT AM I TALKING ABOUT?• Accelerometer is used to measure

proper acceleration.

• Gyroscopes measure the change in orientations.

OBJECTIVES• Develop an inertial Navigation system that can accurately calculate the

motion using a Smartphone.

• Develop and use Kalman Filter to filter data

• Contribute to the ongoing Roller Coaster research project.

METHODS• Use smartphone to take acceleration and gyroscope measurements from

different types of motion

• Take videos of the motion to perform video analysis in order to estimate true position

• Use Matlab to filter data to remove bias and noise in raw acceleration data

• Numerically integrate two times in order to convert acceleration data into velocity and velocity into position data

ANALYSIS PROCESS• Apply force onto a resting cart and record its motion• Filter the data obtained

Raw Acceleration Data Filtered Acceleration Data

• Capture the motion of the cart in a video camera and perform video analysis

• Numerically Integrate filtered acceleration data to get velocity

• Numerically Integrate velocity to obtain position and compare it to true position obtained from video analysis

RESULTS FROM OTHER TYPES OF MOTION

1D Complex motion

2D Simple Motion

2D Incline

Vertical Rotation

Horizontal Rotation

KALMAN FILTER• Recursive system consisting of a measurement update step and

prediction step.

• Uses the equations of motion, known control inputs and

measurements from the sensors to predict and calculate the various variables associated to the system.

• Could not be implemented because of the lack of known control variables in the motions

CONCLUSIONS

• Due to the nature of the motions experimented on, Kalman filter could not be applied onto it.

• The smartphone INS is accurate till 12-15 seconds and quadratic Drift was prevalent after that.

• More research needs to be done with newer smartphones in order to increase accuracy of the data

FUTURE WORK• Experiment on more complex motions like twists and turns

• Use sources of known control variables to attempt the Kalman filter

• Experiment using new smartphones from different vendors

• Develop an INS app for the phone.

• Continue Roller Coaster Research Project

ACKNOWLEDGEMENTS• Randolph College Summer Research Program• Dr. Peter Sheldon• Kacey Meaker ‘08• Alex Tran ‘15• Timothy Slesinger ‘14• Mike Cheng, Virginia Episcopal School• Richard Lin, Virginia Episcopal School