Lab4 Sec3 Team4 Presentation Edited For Blog

Using an acceleromter to explore a virtual world

Background Micro Electro Mechanical Systems (MEMS) Accelerometer

• Principles – Operated under Newton’s 1st Law • F = ma = kx – Under an applied acceleration, the beam deflection is measured and the acceleration is determined

Background Analog Devices ADLX203 Accelerometer

The circuitry in the accelerometer outputs a voltage change  The system has both mechanical microstructures and electrical signal process units

Accelerometers can measure 1,2 or 3 axis  The Analog Devices ADLX203 Accelerometer measures 2 axis ○ Measures both x and y directions (2-axis) with a range of ± 1.7g  Limited to only two axis of rotation and translation

○ It can measure both dynamic acceleration and static acceleration

Prior Research • Virtual Reality – Sega Dreamcast • Used first motion sensing control in the form of a fishing rod – Nintendo Wii • The remote makes use of multi-axis linear acceleration information provided from the ADXL330, an integrated MEMS accelerometer

Prior Research  Use of Inertial Information for Vision  Active vision systems are being used in robotic systems for navigation  Inertial sensors coupled with an active vision system can provide valuable information ○ Use the inertial sensors to find the ground plane  Once you know the ground plane, you can know any objects 3D orientation from that point ○ Use a Virtual Reality Modeling Language (VRML) for describing the 3D world on the computer

Goals and Objectives Hypothesis  The human head and body is always accelerating even as it stands still. Assuming a direct line of sight is maintained only strong enough movements of the head and body would change the focus and person has from an image to another. With the data collected from the MEMS accelerometer, it is possible to measure what kinds of accelerations could affect an image a person is focusing on.

Goals and Objectives • Conduct a simple experiment using a MEMS accelerometer – See if accelerometers can measure movements when placed on glasses • Simulate virtual reality eye-set

– Use MATLAB to analyze the movements of the accelerometer and see if a virtual reality model can be simulated

Lab Methods • Connected MEMS accelerometer to function generator and oscilloscope • Secured accelerometer to eye glasses – X-axis pointed straight out of the individual – Y-axis pointed to the right of the individual

• Measured change in voltage in Oscilloscope and Excel – Calibration results for voltage to acceleration conversion • 1V = 1g • Acceleration normalized based on nominal voltage • Nominal voltage measured by holding accelerometer stationary

Data Analysis  MATLAB  Input data from Excel into MATLAB code  Code converts accelerometer voltage into engineering acceleration units ○ Sensitivity, DC offset, etc.

 Acceleration data filtration ○ Kalman Filtering ○ An efficient recursive filter that estimates the state of a system based on incomplete and noisy measurements

○ Minimum threshold filtering ○ Acceleration data is valid only if threshold is reached

○ Double filter provides consistent and more accurate data

Data Analysis  Kalman Filters  1st Order Implementation ○ Smoothing a noisy sensor input ○ 2 parameters – Amount of previous state information, Dampening effect (R), etc.  The graphs below show that with a Kalman filter implementation, random noise is decreased

R = 0.001

R = 10

R = 100

Data Analysis - No filter

Data Analysis – Kalman Filter

Data Analysis – Double filter

Data Analysis – Double filter

Data Analysis

Data Analysis

Data Analysis

Simulation

This is when you take a look at the video below. Don’t worry, I’ll wait.

Discussion • Virtual Reality Model – Translation in the x and y direction • 1st person perspective • Forward accelerations translate to forward motion based on data analysis

– Roll and pitch of head • Roll and pitch of head translates to roll and pitch of virtual world respectively

– Capable of several viewing angles of virtual world

Discussion • Feasibility – Easily implemented with a wireless connection for gameplay or other applications – Cheap and robust

• Drawbacks – Differentiation of pitching motion from forward acceleration – Limited to 2 axes preventing yawing motion

Conclusion • Virtual reality on a rudimentary level – Benefits – Cheap and robust – Easily implemented – Can be built with a simple microcontroller and 2 3-axis accelerometers for redundancy – Additional improvements may involve integrating a camera for verifying actions