HCI on WANG Zhaofeng · Course Portfolio

HCI Lab: Streaming Chat in Unity with the DeepSeek API

Mon, 19 Oct 2026 00:00:00 +0000

The lab asks us to build a chat character in Unity, then study how people experience the interaction. I replaced the local LLMUnity model with the DeepSeek API to avoid downloading and running a model on this PC. The chat still streams its answer and locks the input until the reply finishes.

This changes the lab’s offline setup: internet access and API credit are required, and the measured delay includes the network and provider. It is an API-based adaptation. The course deck specifies Unity 6 with URP; this UI-only project uses Unity 6000.0.23f1, the Built-in Render Pipeline and TextMeshPro. Acceptance of these changes is for the instructor to confirm.

HCI Lab 2: Touch Controls and an Android Build

Fri, 16 Oct 2026 00:00:00 +0000

The desktop learning task has two roles: the cyan sphere follows its neural policy, while the gold target can move. For the mobile version, a touch pad replaces keyboard control of the target. It does not directly steer the sphere.

Windows Editor preview of the touch layout. This is not an Android screenshot.

HCI Lab 1: Machine Learning in Unity with ML-Agents

Mon, 12 Oct 2026 00:00:00 +0000

Signal Chase starts with a simple task: move a cyan sphere towards a gold target. The extension makes the target move and reflect from the platform edges. Both tasks use learned PPO policies, not a scripted chasing rule.

Runtime camera capture of the moving-target task. The desktop control overlay is not included in this capture.

HCI Lab: Gaussian Splatting from Images to Unity

Sat, 10 Oct 2026 00:00:00 +0000

This lab connects image-based training to a Unity scene. The trained model stores coloured, transparent 3D Gaussians rather than a triangle mesh. Each Gaussian has a position, scale, rotation and opacity; overlapping splats form the image seen by the camera.

Actual Unity camera capture, with the desktop control overlay excluded. Soft edges and stretched details are still visible.

HCI Homework 4: Phone Shapes and Everyday MR

Thu, 08 Oct 2026 00:00:00 +0000

Lecture 10 - The shape of smartphones Link to heading

Earlier keyboard-based smartphones reserved much of their front surface for typing. Touchscreens gave that space to the display: the same surface could become a keyboard, a map, or a video player. Larger screens suited reading and media, but also made one-handed use harder.

Changes in input and screen space. The last panel is a proposal, not a product forecast.

HCI Homework 3: An AR Guide for Finding a Room

Sun, 04 Oct 2026 00:00:00 +0000

Lecture 7 - Three ideas for augmenting human abilities Link to heading

I focused on small difficulties that AR could help with: finding a place, understanding language, and remembering a sequence of actions.

Concept sketches. The backgrounds and overlays are schematic.

Room Guide. Show the next direction near a corridor junction, with the destination name always visible. It could reduce the effort of matching a floor plan to the place in front of you. Wrong or outdated positioning would be its main weakness.

HCI Homework 2: Body Interfaces and the Ultimate Display

Wed, 30 Sep 2026 00:00:00 +0000

Lecture 4 - An HCI researcher: Pedro Lopes Link to heading

Pedro Lopes leads the Human Computer Integration Lab at the University of Chicago. His work studies interfaces that connect to the body through touch, muscle stimulation, and other sensations. The computer can provide physical feedback instead of only showing instructions on a screen.

Two recent projects make this approach quite concrete:

Generative Muscle Stimulation, CHI 2026. The system uses a person’s view and body pose to generate physical assistance for everyday tasks. It constrains the generated instructions using joint limits and knowledge about muscle stimulation. This makes context important: the same object can require different actions in different situations.
Myo Action, CHI 2026. Muscle signals trigger stimulation that assists an action the person has already started. The aim is to speed up movement while preserving the person’s decision to act.

The interesting HCI question is control. Assistance should be easy to interrupt, and the person should understand when the system is acting. A technically effective movement could still feel wrong if it happens unexpectedly.

HCI Homework 1: Affordances, Gestalt Laws, and Dark Patterns

Sat, 26 Sep 2026 00:00:00 +0000

This homework looks at three ways design affects behaviour: the actions an object suggests, the visual groups people perceive, and the choices an interface tries to steer.

Lecture 1 - Affordances Link to heading

An affordance is an action that an object makes possible. Its shape and signs help people discover that action. A well-designed door needs little explanation.

Good example: a crash bar on an exit door Link to heading

Figure 1. A horizontal crash bar suggests “push here” through its size, position, and direction.

HCI Lab 1b: Unity Setup with an Existing Project

Tue, 22 Sep 2026 00:00:00 +0000

Unity was already working for my Computer Graphics project, Corebound. For this lab, I am documenting that existing environment and its recorded results instead of creating another empty project. The screenshots below come from the Windows player run on 11 October 2026.

The separate ML-Agents lab adds trained reaching and chasing policies. This page remains the basic Unity setup record.

Environment and project structure Link to heading

Corebound uses Unity 6000.0.23f1, the Built-in Render Pipeline, and a Windows build with the Mono scripting backend. It is a small 3D game: move a sphere, collect eight energy cores, and reach the reactor.

HCI Lab 1a: Setting Up the Course Blog

Fri, 18 Sep 2026 00:00:00 +0000

The lab asks for an online blog and a record of the setup process. My blog uses Hugo Extended with the hugo-coder theme. It is a course portfolio rather than a single stream of unrelated posts.

Organising the work Link to heading

The Courses page gives each subject its own space. Posts carry a course code, such as HCI, so they appear automatically on the right course page. Within each homework, lecture headings keep the answers separate. Lab work gets its own post.