May 2019

MAXAR

To give an impressive presentation, we always use visual aids to emphasize key points. However, we usually use presentation slides. Then our team came up with a question – Why don’t we combine AR technology with presentations? MAXAR is a mobile application which utilizes AR technology to facilitate presentations. Users prepare their visual aids in our application by adding 3D elements or 3D text. They can either record their presentations and upload them to the Internet or deliver their presentations directly in front of the audience using their own device.
Our application shows an incredible potential in STEM education, e-learning, and entertainment.

SpaceKey: Exploring patterns in spatial databases for property searching

The demand for property is steadily rising with the significant market size of real estates across the world. Looking at the existing property searching applications, they only provide simple filtering functionalities about the properties, such as price, size, etc. However, in reality, users might have some complex requirements about the surroundings of their desired properties. To deal with this unsatisfied demand, we choose to facilitate an advanced query type – Spatial Pattern Matching (SPM), which is a new type of spatial group keyword query raised in recent research. Led by our supervisor, Dr. Reynold C. K. Cheng, its solutions have been well-discussed in theory. In our project, we want to fit these theories in this real-life scenario by resolving practical issues, as well as realizing some additional functionalities related to property searching. In the end, we have developed a standalone web-based application to demonstrate the importance and practicality of this newly proposed functionality.

Understanding financial reports using natural language processing

This project investigates how mutual funds leverage credit derivative by studying their routine filings to the U.S. Securities and Exchange Commission. Credit derivatives are used to transfer credit risk related to an underlying entity from one party to another without transferring the actual underlying entity. Therefore, we developed both rule-based and NLP based methods to extract CDS information from thousands of files and aggregate into a dataset and further developed a one of its kind search engine to provide quick access to historic data

Evolving Human-Like Micromanagement in StarCraft II with NeuroEvolution and Reinforcement Learning

In our project, we implement neuro-evolution using NEAT and reinforcement learning using Sarsa(λ) on micromanagement scenarios in StarCraft II involving the small-scale precise control of combat units. Using our developed training framework for applying NEAT to StarCraft II, we evolved neuroevolutionary agents that learned to demonstrate precise hit-and-run strategies to beat the in-game AI in ranged vs melee matchups. Our reinforcement learning agents using Sarsa(λ) learned to be successful in more complex micromanagement scenarios involving enemy engagement selection and timing. Our results serve as a proof-of-concept of the benefits and potential of the applications of these techniques in video games and represent meaningful contributions to the wider video gaming and artificial intelligence communities.

A MMORPG to raise public awareness on Computer Crimes

Create a Massive Multiplayer Online Role-Playing Game (MMORPG) populated with computer mechanism related elements that can help players raise awareness on cyber-secure behaviors in their daily computer and smart device usage. There are two game modes available for demonstration, namely single player career and multiplayer arena. The former displays how the story background is set and links the logic flow between the player and the AR device that allows the player to visualize computer mechanisms and cyber behaviors into magical spells and abilities. The latter allows up to 20 players to practice these abilities in a simulated battle arena where the objective is to take down the enemy “server”, which can then disable enemies to respawn from their server data and obtain victory.

Piezoelectric shoe

Harvesting mechanical energy from human motion in day to day life is a very attractive way of creating sustainable electric energy and a piezoelectric shoe exactly does that. Since it is a wearable energy harvester, it is very convenient to use it and has many advantages which will be discussed further. It generates electricity through piezoelectric effect in the shoe and stores it in a battery which can further be used to charge your phone. When a person walks electrical energy is produced and that energy is stored in the battery.

Smart pet keeping system

This smart pet keeping system includes multiple modules such as neck device, cameras, motion trackers and pet toilets. The system has three main features, an entertainment system (Remote control toy), a monitoring system (Danger Alert) and a health care system (Health Tracking). It aims to help saving time on pet keeping, keep pets healthy, keep a close relationship between pets and pet owners and put an end to abandon pets.

Smart life vest

The Smart Life Vest which aims at increasing the efficiency of rescue operations in water. The primary features of the Smart Life Vest include sending automatic SOS signals containing the wearer’s GPS location to the nearest rescue station using radio or microwave signal, which will immediately alert the rescue party of an emergency. The Smart Life Vest will also produce its own power using a built-in wave generator and make use of chemicals that react with water to facilitate its auto-inflation mechanism. Both of these will increase the shelf-life of the Smart Life Vest and reduce the maintenance cost.

Find A Seat App

Many students endured a difficult and time-consuming search for an empty chair, particularly during the examination period. Furthermore, hogging seats with personal belongings for a long time are some notorious acts that are very common in the libraries. With a view to tackle these issues, this project “Find A Seat App” aims at building an application that can detect the occupancies of the seats in HKU Main Library in real-time. Two approaches, i.e. using Computer Vision techniques and Internet of Things (IoT) device, are examined and discussed in the project.