Trail Sense

TrailSense is a hands-free nature companion designed for Meta display glasses. It helps hikers in Southern California stay safe and stay curious by providing quick, glanceable plant identification and gentle safety guidance — without stopping the flow of the trail.


01

Why TrailSense?

Hikers are often curious about the plants they see.

But taking out a phone interrupts the pace of the trail and breaks attention.

TrailSense provides plant identification in place, without stopping.

Quick. Glanceable. Heads-up.

02

Designed for the Curious Hiker

TrailSense is made for hikers who want to understand the landscape while staying in motion.

Context Southern California + Sierra trails

Behavior Notices plants, prefers not to stop often

Need Identification that doesn’t disrupt movement

Safety Wants to avoid toxic look-alikes

Hands Often full (poles, water, scrambling)


03

Scope

This project focuses on the user experience, not the recognition model.

In Scope

  • HUD visual design

  • Curiosity Mode

  • Safety Mode

  • Low-Confidence Mode

  • Confidence dot system

  • Voice and gaze interaction flow

  • Field testing through Wizard-of-Oz

Out of Scope

  • Training the computer vision model

  • Building Meta glasses hardware

  • Identifying all plants everywhere


04

Curiosity Mode

05

Medium Confidence

06

Low Confidence

When TrailSense is uncertain, it does not guess.

Instead, it presents both possibilities and offers one distinguishing feature the hiker can check visually.

This approach prevents false confidence and builds trust through transparency.

07

Safety Mode

Safety color is reserved ONLY for risks so the user associates orange with “pay attention” consistently.

07

Why Confidence Needs to Be Visual

I drew on interface patterns from aviation HUDs, Tesla Autopilot, and Apple’s spatial interaction design, where shape and opacity, not text, communicate confidence. This allows TrailSense to express certainty in under one second, without distracting hikers or reducing situational awareness.

TrailSense communicates certainty using a simple confidence dot that pairs a visual signal with tone.

Solid = confident

Dimmed = likely

Hollow = uncertain

Instead of making the user read or interpret a sentence, the system uses repeated visual + audio pairing, allowing hikers to understand confidence instantly, without breaking flow or looking away from the trail.

08

Why Scope Matters

I intentionally scoped TrailSense to a small set of plants that hikers frequently encounter in Sierra wet-meadow environments.

The goal was not to identify everything, but to support recognition, safety awareness, and learning through meaningful patterns.

The core learning moment comes from look-alike species, where TrailSense uses a hollow confidence dot and offers one distinguishing cue to help the hiker verify visually.

This prevents false certainty and makes the system feel collaborative rather than prescriptive.

09

What’s Next

TrailSense is the first version of a simple, glanceable plant ID system.
There are several areas to improve:

1. Location-Based Suggestions

Show plants that are likely in the area based on GPS and habitat.

2. Better Look-Alike Detection

Use stronger visual comparison to warn hikers when two plants are commonly mistaken for each other.

3. Personalized Learning

Remember what the hiker has identified before and adjust suggestions or difficulty over time.

4. Multi-Sensory Guidance

Add small haptic taps or short chimes for confidence and safety cues when voice isn’t ideal.

5. Expanded Species Coverage (Scoped by Region)

Slowly add more plants for additional regions while keeping the experience simple and not overwhelming.

6. Offline Mode

Allow identification and safety guidance even with no cell service.

©

Russell Peng

2025