kit

Undead Unalive is a fun to play first person 3D survival horror game. The game is set in a dystopian future where Kobid-19 (covid spoof) has taken over the world and unleashed an outbreak of blood thirsty zombies. Further, to incorporate elements of covid into the game, we replaced guns with vaccines, and included face masks that would allow the player to dodge the cough particles exhaled by the zombies. The game implements a huge map with multiple zombie zones. Additionally, we added cutscene animations at the start of the game, implemented user interfaces, score multipliers and audio to create an immersive experience for the player.

Cellsverse is a multiplayer game. It is of best of 3 format. We provide three maps for players, namely lung, liver and heart. Players can increase their strength by eating the nutrients that are randomly spawned on the map. Players can attack their opponents by either gun shooting or melee. They can also use different abilities which costs mana. When the hit point decreases to zero, the player dies and the game proceeds to the next map. Each map have their own features and map design to mimic the actual human organ. Through this game, we hope players can immerse into the game world, and knows more about the battles between the immune system and the virus that happen in their body every day.

Uni Fighters is a game that we have made for the course COMP3329 – Computer Game Design and Programming. It is a 2.5D fighter game set at the University of Hong Kong, portraying a student who gets into a fight for bumping into his schoolmate. Players will play as a student to fight against the opponent using combat mechanics. We created all the models by ourselves in Blender, including the characters and the map, as well as other assets such as shaders and particle effects using tools like Unity’s Shader Graph and VFX Graphs.

Dynamic state representation learning that can accurately describe dynamics can significantly accelerate reinforcement learning training. However, deformable objects have very complicated dynamics and high DoFs. We propose DiffSRL, an end-to-end dynamic state representation learning pipeline that uses a differentiable physics engine to learn the representation of deformable objects. Our specially designed loss function can guide neural networks to be aware of dynamics and constraints. We benchmark the performance of our methods as well as other state representation algorithms with downstream tasks on PlasticineLab. Our model demonstrates superior performance most of the time on all tasks. We also demonstrate our model’s performance in a real-world setting with two manipulation tasks on a UR-5 robot arm.

Supply chains containing complicated networks of producers, transporters and consumers have played an integral part in the expansion of online and offline businesses worldwide. With so many stakeholders, Supply chain visibility which is the ability to track different goods at each point has become essential. With increased visibility, businesses can optimise their supply chain while consumers can ensure the ethicality of their sourcing practices. However due to inefficient solutions, business owners report poor visibility facing 20% loss in goods annually, whereas most consumers would be willing to pay an extra 2-10% for traced products. Goodchain is a platform consisting of web & mobile applications which provides transparency and traceability in the supply chain through the use of NFT and blockchain. By solving the inefficiencies while catering to all stakeholders in the industry, Goodchain aims to disrupt the supply chain management industry and add immense value to all its members.

In this project, I made a robot dog for STEM education based on 3D printing and Arduino development kit. On top of this, I completed the integration of ESP32 and MPU6050, which gave the robot the ability of motion detection. Through the calculation of the motion data and the implemented PID control system, I completed a very good Self-Balance function. In addition, I also tried to apply the self-balance algorithm to the motion state which improves the motion posture and optimize the performance of the robot dog walking on certain terrain. Finally, I sorted out the problems encountered in the development process and some important principles into a document for STEM education, which can help others better understand and apply related content.