IRIS Research Staff

SIJIE YU
MS Student

GPS/IMU Navigation

Navigation is useful in many applications, may be robotics path planning, automatic vehicle localization, and mobile mapping of large urban architectures. Navigation can be archived by fusing spatial data from multi-sensors, such as GPS and IMU (inertial measurement unit). These two sensors work independently to output 3D spatial information. More precisely, GPS measures x,y and z values; while IMU measures roll, pitch and yaw Euler angles. Combining their data together, we can have full 6 DoF (degree of freedom) of our motion. However GPS suffers "loss of sight" problem due to the blocking of buildings, trees and even bad weather. Besides GPS accuracy is normally at an order of sub-meter. IMU suffers significant drift error due to double integration of time. Therefore their data are usually fused together to generate accurate spatial information. For this purpose, a C++ application is implemented to control these two sensors and download spatial data in real-time. Then post-processing is performed to De-noising, Synchronizing and Combining their data. That is

. Fusing GPS 3D positions (x,y,z) and IMU acceleration using Kalman filter; to provide precise and stable 3D positions and 3D velocities,

. Since both equipments work in independent working rates, synchronization is realized using time stamps.

. For integration, data are saved into separate small files for post-processing.

Sensor Controlling Program (C++)	Trajectory reconstructed from GPS Data

Full dimensional road surface is very useful for many applications, including battle field simulation, pavement inspection, and robotics navigation. Therefore fast profiling road surface has been a great research topic ever since 80's. However all the previous approaches lack of efficiency or can't obtain 3D data directly. In this project I use a laser, a DGPS and an IMU to generate 3D road surface in a fast and accurate mode. For this purpose, a multi-sensor integrated data acquisition system is instrumented. Then post-processing is performed for De-noising, Integration, Interpolation and Reconstruction. The final output is an eye-appeasing and precise description of the surface topologies.

Implementation of an Interactive and Composite Web 3D Maps

This project is to implement an Internet based interactive 3D geologic map, which integrate multiple layers of 2D maps and 2.5D elevation map. To provide interaction with customer, standard VRML scripts are fixed to include Blaxxun Contact nodes extensions. Basic navigation (e.g. zoom in/out, move right/left) are provided to view the 3D world. http://imaging.utk.edu/~sijie/overlayswitch/overlayswitch.html (To view the prototype, a Blaxxun Contact viewer must be preinstalled, http://www.blaxxun.com/en/products/contact/eula.html)