Page 92 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Electronic & Computer Engineering

to determine a robot’s position through wifi data. In the first exploration, it is discovered that the traditional mean and
variance model of the weight distribution provides a higher accuracy than applying linear regressions in the
situation. The second exploration implied that machine learning can be very effective in determining the position of the
car within an acceptable range of error with limited data.

Supervisor: SONG Shenghui / ECE
Student: DONG Tian / ELEC Course: UROP1100, Fall
LUK Wing / ELEC UROP1100, Fall
JIN Menghe / ELEC UROP2100, Fall

In the recent years, autonomous driving is becoming the main trend in the market and is applied to more and more
areas. In autonomous driving, sensor is an indispensable component to extract the information from the surrounding
environment. In this project, we try to enhance the existing sensor system of the Kobuki turtlebot and make it possible
to detect the transparent objects. We did several experiments based on the result we got in the summer UROP and
understood more about the sensor. However, at this stage, the problem of sensor accuracy and sensor frequency still
need further improvement.

Supervisor: SONG Shenghui / ECE
Student: KANG Yanming / COMP Course: UROP1100, Summer
WAN Yuxuan / SENG UROP1000, Summer

This project is aimed to use machine learning to discover the minimum time needed for a robot car to travel one lap on a
certain track. We use the Arduino Uno chip to guide the car and to set the speed, as well as to collect the data. We use
Tensorflow (an open-source machine learning library) to process the data. The task is split into two parts. First, a model
is trained to predict whether a given set of speed will make the car go out of the track. Second, another model is trained
to predict the travel time and find the shortest time without leaving the track.

Fabrication and Characterization of a 2000-Atmosphere Pressure Sensor

Supervisor: WONG Man / ECE
Student: TANG Zili / ELEC Course: UROP1100, Fall
UROP2100, Spring
UROP3100, Summer

Many current applications, ranging from oil drilling down the ocean to vehicles on the road, require precise
measurement of high pressure. Therefore pressure sensors that can perform under high pressure and harsh
environment are a popular research topic. The high pressure sensor that I am working on is a tiny bulk-type device made
of n-type silicon that includes a device part and a cap part. There are piezoresistors on the pressure sensor device part
whose resistance will change under different hydrostatic pressure loads, resulting in different voltage outputs that
indicate pressure. Finite element simulations of the sensor will be presented below.

Nanofluidic Chips for Biomedical Applications

Supervisor: YOBAS Levent / ECE
Student: LEUNG Ziuwin / BIOT Course: UROP1100, Spring

82