Legibility by design ft. Ryan Powell (Waymo) | Config 2026

Summary of Technological Concepts & Product/UX Features (Auto-generated subtitles)

Waymo’s mission framing & scale

• Ryan Powell (Waymo design lead) argues autonomous driving is not a “cool novelty” but a public health imperative, citing an open letter from the medical community about road deaths.
• Waymo shared operational metrics:
  • 200M+ rider-only miles
  • 20M+ passenger trips
  • 500k+ rides per week across 11 US cities
• Planned expansion includes London and Tokyo.

Core design problem: statistical safety vs emotional trust

• Even when a system is objectively safe, trust is emotional.
• Riders don’t think in crash statistics; they need:
  • comfort
  • convenience
  • clear intent
• Safety framing mentioned in the talk:
  • 92% reduction in serious injury crashes vs human drivers (as stated)

Shift away from driver-assist to Level 4 autonomy

• Early research used a Lexus fleet with driver assist (human takeover required).
• Findings: users over-trusted the system when it behaved “reasonably well.”
• Observed behaviors during testing:
  • Riders became distracted at highway speeds (e.g., texting/makeup)
  • Some riders even nodded off
• Product strategy consequence:
  • Fully commit to Level 4 autonomy where the passenger never intervenes
  • Waymo began “all-in autonomy” in late 2019 in Phoenix

Trust design grounded in communication cues (proxy for missing human cues)

• Without a human driver, Waymo must replace the “unspoken dialogue” of human driving—especially non-verbal cues (e.g., body language/eye contact equivalents).
• Designers aim to create a “new language of trust” so the vehicle’s behavior feels legible and confidence-building.

Curated passenger UI: “ruthless curation over complexity”

• Instead of showing raw sensor output, Waymo designs UI to show what the car can perceive without overwhelming riders.

3D scene rendering approach (passenger display)

• Grounded road surface with highlighted elements:
  • crosswalks
  • road markings
• The vehicle’s route shown as a predictable green path
• Nearby vehicles as simple, stable 3D geometry
• Pedestrians/cyclists encoded with distinctive visual treatment:
  • pedestrians: light blue glow
  • cyclists: green underlay
• For pedestrians/cyclists, Waymo uses a hybrid approach:
  • layered “real-time laser points” over figures
  • motion feels more embodied and less abstract

Context layering

• Buildings provide geographic context but are dimmed so they don’t compete with higher-importance elements (nearby actors).

High-yield UI detail: real-time orange traffic cones

• A frequently cited success: rendering orange traffic cones during construction zones.
• Research finding:
  • when riders see cones matching the real world, they instantly relax
  • the detour/obstacle becomes understood rather than confusing

Status layer for direct, situation-specific explanation

• UI includes an overlay explaining what the car is doing, such as:
  • slowing in a school zone
  • waiting for an intersection to clear before turning
• Intended use:
  • more “direct” when rider attention is needed.

Sound design using psychoacoustics (musical key of E)

• Waymo anchors an auditory hierarchy in the musical key of E, associated (per the speaker) with joy and clarity.
• Example sound-to-action mappings:
  • first door open: soft “spa music” to transition from outside world to the car
  • start trip: rising scale chime
  • arrival: falling scale chime (“You’re here…”)
  • higher-attention moments (e.g., closing door): sounds shift away from the base key to create “mild polite tension”
• Design rationale:
  • avoid annoying constant alerts
  • use sound sparingly and intentionally

External communication for other road users (city-scale legibility)

• After expanding from Phoenix to San Francisco, the system needed to handle far more pedestrians:
  • 15x in neighborhoods
  • 50x in the financial district
• Goal: compensate for missing human-driver signaling in micro-interactions, such as:
  • right-of-way decisions
  • “who goes first?”
• Proposed proxy for human intent:
  • motion-based cues
  • example: a deliberate “nudge forward” at four-way stops to signal planned turn/priority

360° Lidar dome as an external display

• When the car is stationary (e.g., yielding to pedestrians), Waymo uses the lidar dome as a 360° external display.
• Initial operational use:
  • rider initials to identify the correct vehicle
• Expanded intent signaling uses three principles (as described):
  • legible
  • easy to learn
  • glanceable

Example behavior at stop signs

• An external soft forward pulse indicates: “I see you. Go ahead.”
• A targeted signal can also broadcast intent to vehicles behind.

Safety beyond the cabin: “dooring” prevention

• Since the vehicle can see 360°, it mitigates door-related hazards (“dooring”).
• Exit-flow safety buffer includes:
  • cabin alerts with a directional chime
  • an on-screen graphic
  • a gentle audio reminder to look before exiting
• Example subtitle callout:
  • “Scooter approaching.”

Extending trust to special user contexts (Waymo Teen Accounts)

• In 2021, Waymo launched teen accounts (ages 14–17) as a responsibility requiring UX for two users:
  • teen riders: designed for freedom and a stranger-free environment
  • parents: reassurance features such as:
    • real-time location tracking
    • easy connection to rider support
• Central message:
  • successful trust-building makes the interface feel like a “silent partner,” not just a control surface.

Future direction: conversational autonomy

• Concept: shift from app/map interactions to natural conversational control—like telling a human driver where to pull up.
• Goal: achieve “perfect legibility” so the UX becomes invisible and conversational, ultimately returning freedom to passengers.

---

Main speakers / sources

• Speaker: Ryan Powell — Lead of Design, Waymo
• Source referenced: “medical community published an open letter” (unnamed) about America’s road deaths as a public health crisis.