Robotics Engineer for Voice AI Integration

Robotics Engineer Needed: Voice AI Integration for Unitree G1 Humanoid — London Trade Show (June 6-10) We're a communications company bringing our Unitree G1 EDU Ultimate humanoid robot ("Bob") to a London trade show. We need a UK-based robotics engineer to integrate our Voice AI API and program interactive booth behaviors. Part-time, short-term. The core job: Connect the robot to our Telnyx Voice AI API so attendees can have real-time conversations with it — someone speaks, Bob listens via mic, our API generates a response, Bob speaks it back through his speaker while gesturing naturally. We handle: The Voice AI backend, conversation scripts, booth/stand design. You handle: Robot-side audio integration, gesture programming, safety, and show-floor reliability. ─── 🎯 What You'll Build 1. Voice AI Integration (TOP PRIORITY) • Connect to our Telnyx Voice AI API (WebSocket-based real-time audio streaming) • Flow: attendee speaks → audio streams to Telnyx → AI response → plays through robot speaker • Robot triggers matching physical gestures (wave when greeting, point toward demo, etc.) 2. Physical Interaction & Movement • Hand/arm motions to safely hand out flyers, stickers, small swag • Gesture library: wave, point, thumbs-up, beckon • Idle "attract mode" — subtle motion loops that draw attention • Greeting sequence on approach 3. Show-Floor Awareness • Spatial awareness / obstacle avoidance via onboard LIDAR • Maintain position within booth boundary • Rotate and face inbound foot traffic 4. Trigger-Based Behaviors • Proximity: wave or speak when someone gets close • Touch: respond when someone taps its hand (tactile sensors built in) • Voice: turn toward "hello" or "Telnyx" 5. Safety (NON-NEGOTIABLE) • Motion limits to prevent sudden/unsafe movements • Auto-freeze on unexpected resistance • Quick disable toggle for booth staff • Stays within booth area — no wandering 6. Trade-Show Reliability • Auto-restart on crash • Battery management (~2 hrs active, spare battery included) • Must work on event WiFi (dedicated hotspot as backup) ─── 🚫 What You Do NOT Need to Do • Build the Voice AI flow (our team handles API side) • Design the stand or headphone station • Create conversation scripts ─── ⚠️ Known Issues (from our internal testing) • No spatial awareness out of the box — executes commands regardless of obstacles • SSH connection refused when connecting directly • USB data ports under back cover not yet accessed • Mode switching requires stopping current mode first • Mobile app only via Beta channel ─── 🤖 Robot Specs — G1 EDU Ultimate • 43 DOF total (arms: 7 DOF each, waist: 3 DOF) — very expressive • Dex3-1 force-controlled 3-finger hands with tactile sensors • 3kg max arm payload — handles flyers/swag easily • 100 TOPS onboard compute • 9000mAh battery + spare (~2 hrs active) • High + low level SDK access Technical Stack • SDK: Unitree SDK2 (Python) + CycloneDDS • Voice API: Telnyx Voice AI (WebSocket) ─── Timeline • Development: June 6-7 in London • Tested and stable: by 7 AM June 8 • Live at show: June 8-10 • On-site in London strongly preferred Location: Must be UK-based with access to the robot