10-02-2012, 03:00 PM
Final Demonstration: Dead Reckoning System for Mobile Robots
Problem Statement
Dead reckoning is navigation based on measurements of distance traveled from a known point.
Use a mobile robot and develop a synthesized dead reckoning navigation algorithm.
We will integrate various sensors.
Robot and Sensors
MARK III Robot with OOPic Chip
MEMS Accelerometer
MEMS Gyroscope
Shaft Encoders
Digital Compass
Algorithms We Tested
North Bound using Compass
Turning using Gyroscope
Turning using Encoders
Turning using Compass
Encoder x,y Movement
Algorithms We Tested (Cont)
Accelerometer x,y Movement
One-Direction using Accelerometer and Encoders
Turning using Encoders and Gyroscope
X, Y Path integrating sensors
Z Path using all sensors
Merging Data
Accelerometer and Encoders data merged for translations
Gyroscope and Encoders data merged for rotations
Weights found for each sensor by calculating percent errors
Merging Data (Cont)
Weights
Gyroscope - Rotational
CW – .11
CCW – .19
Accelerometer - Translational
.12
Encoder
Rotational
CW – .89
CCW – .81
Translational
.88
DEMO 1:
North Bound Using Compass
Robot will turn and travel towards north where ever it is initially pointing
DEMO 2:
Z-Path integrating sensors
Uses combination of all sensors
Fusion of sensors
ORANGE – gyroscope turns 90 degrees, CW
RED – accelerometer travels 56 cm
BLUE – encoder turns 150 degrees, CCW
GREEN – encoder travels 64 cm
BLACK – gyroscope and encoder merged to turn 150 degrees, CW
PURPLE – accelerometer and encoder merged to travel 56 cm
Problems With Each Sensor
Accelerometer
- Converting values
- Unable to use Digital signal
Encoder
- Mounting on Robot in an aesthetic manner
Gyroscope
- Analog signal sensitive to noise
- Converting values
Compass
- Accuracy is very dependent on environment
