Rethinking the Touchpad — Akshay R Nair

Modern laptop touchpads are sophisticated spatial sensors capable of tracking raw absolute coordinates, contact IDs, and multi-touch surface assignments. Yet, standard operating system drivers reduce this hardware to a relative mouse surrogate, collapsing high-resolution spatial inputs into simple cursor deltas and as few hardcoded OS swipes.

This project explores an alternative interaction model that treats the touchpad surface as a primary, absolute two-dimensional surface. By mapping absolute coordinate zones, movement vectors, and dwell times, we built a semantic gesture vocabulary that enables deep, continuous interactions, like reading long-form documents, without requiring cursor manipulation or context-switching to the keyboard.

We expected a gesture shortcut. We found an absolute plane.

When we started investigating laptop trackpads, our initial focus was on the disconnect between what hardware sensors expose and what operating systems deliver. Standard drivers treat modern multi-touch trackpads as mouse-replacement relative pointers, discarding coordinate resolution in favor of cursor acceleration.

Once we mapped the raw Linux evdev interface, the actual interaction breakdown hit us: during long-form reading, users face constant micro-friction. Every page turn, link jump, or panel toggle forces a context switch—either moving the cursor precisely to a tiny UI target or lifting hands to hit multi-key shortcuts.

THE SOLUTION

We Built an Absolute Spatial Interaction Pipeline

Instead of accepting relative cursor movement as the default paradigm, we built an integrated system that treats the touchpad surface as a normalized 2D unit square ($[0,1] \times [0,1]$). By mapping absolute coordinate zones, directional vectors, and dwell thresholds, we established a learnable gesture vocabulary that controls web reading directly without cursor intervention.

Zero-Cursor Document Control: Swiping on designated spatial zones executes logical document actions—vertical swipes on the right edge scroll, horizontal swipes on the bottom edge flip pages, and left-edge swipes toggle peek panels.
Declarative Gesture Architecture: All spatial rules are defined as data inside actions_repo.py, allowing gesture thresholds and mappings to be modified instantly without touching core pipeline logic.
Decoupled Three-Layer Engine: A Python evdev pipeline normalizes raw hardware inputs, classifies gestures, and emits simulated keyboard events to drive a web reader cleanly without requiring custom inter-process communication.

High-Fidelity Validation & Debugging

To validate touch precision and debug coordinate boundaries, we developed touch_actions.py—a Pygame visualizer rendering live finger movement, active zone triggers, and normalized coordinates in real time.

Live spatial gesture tracking demonstration on document navigation.

Raw coordinate event parsing and real-time action mapping.

Demonstration Video

Here are the demonstration videos showcasing our Touchpad prototype in action—navigating long-form documents through zone-based spatial gestures without cursor manipulation:

Full-screen demonstration of continuous document navigation and spatial reading gestures.

Real-time interaction view showing physical touchpad contact tracking and gesture trigger mapping.

USER TESTING

What We Learned from User Testing

We evaluated the complete prototype through structured reading tasks where participants navigated multi-page documents, followed embedded links, and managed bookmarks. The qualitative evaluation surfaced key behavioral insights alongside hardware usability bounds:

Spatial Reasoning Over Memorization: Users rapidly internalized the surface as a physical map. Participants naturally described “the right side for scrolling” or “the top corner for bookmarks,” proving that spatial congruence allows users to deduce gestures rather than memorize rules.

Zone Boundary Rigidity: Hard coordinate cutoffs emerged as the main operational breakdown. Inherent human touch imprecision caused unintended activations near corner boundaries, highlighting the need for fuzzy or probabilistic zone classification in future iterations.

Testing Insights and Impact

Spatial Reasoning Over Memorization: Users internalized the touchpad as a physical map within 3 minutes. Participants naturally deduced gestures—like “right edge for scrolling”—rather than memorizing arbitrary multi-finger combinations.
Zone Boundary Rigidity: Hard coordinate cutoffs were the primary operational failure point. Natural touch imprecision triggered accidental inputs near corners, pointing to a clear need for probabilistic, fuzzy boundary classification in future iterations.
Zero-Cursor Fluidity: Participants navigated multi-page documents, flipped pages, toggled peek panels, and set bookmarks using spatial surface zones alone—without touching the cursor or keyboard once.
High-Signal Feedback: Users consistently rated spatially grounded absolute zones as significantly more intuitive and natural than standard OS multi-touch swipes.

Why It Matters

Standard OS gestures (like two- or three-finger swipes) lack semantic logic—they act more like hidden easter eggs than natural controls. Operating in absolute coordinates transforms a single finger into an intuitive 2D interface.

To prevent accidental triggers during normal cursor work, a dedicated hardware toggle (e.g., a function key) flips the touchpad into a dedicated Reading Mode.