How Meituan’s AI‑Powered Smart Helmet Boosts Delivery Rider Safety

01. Introduction

Meituan’s smart helmet, designed for food‑delivery riders, integrates Bluetooth calling, helmet‑wear detection, an AI voice assistant, and collision‑fall monitoring. The core software focuses on combining active and passive safety to protect riders effectively.

02. Safety Risks and System

2.1 Safety Challenges

Riders face unique safety challenges: high‑pressure peak periods, complex traffic, vehicle instability (motorcycles, e‑bikes), adverse weather, fatigue, and risky behaviors linked to work pressure.

Accidents and violations: Pressure leads to distracted and dangerous driving, increasing crash risk.

Vehicle types: Motorcycles and e‑bikes are inherently unstable and lack sufficient protection.

Work environment: Harsh weather and irregular rest further raise risk.

Behavioral factors: Poor safety awareness and knowledge contribute to unsafe riding.

2.2 Safety System

Rider safety protection comprises active safety and passive safety , forming a complete safety system.

Active safety prevents accidents through technology and behavior guidelines, such as obeying traffic rules and using systems like ABS and traction control. AI‑driven visual assistance can detect tilting or abnormal motion to issue early warnings.

Passive safety minimizes injury after a crash using protective gear (helmets, gloves, pads) and, where possible, vehicle‑level passive safety designs.

Both aspects complement each other, providing comprehensive protection for riders.

03. Smart Helmet Safety Capabilities

3.1 Active Safety

1) Clear two‑way communication at 60 km/h – High speed creates wind noise and echo that degrade call quality. Meituan’s hardware team solved this with physical noise reduction, beamforming, AGC, and AEC algorithms, ensuring clear calls even at 60 km/h.

Acoustic testing with a realistic head model and microphone array confirmed stable performance under wind and urban noise.

2) AI voice assistant – Riders need to accept orders, navigate, and reply to messages while riding. Voice interaction frees their hands and eyes. The helmet’s wake‑word “Xiao Ling Xiao Ling” activates a large‑model AI assistant that provides order decisions, route planning, emotional support, and one‑click 120 emergency calls.

“Xiao Ling Xiao Ling, help me pick up a nearby order.”

“Xiao Ling Xiao Ling, my orders are too few, what should I do?”

Decision empowerment: Real‑time order heatmap analysis and optimal routing.

Emotional interaction: Detect anxiety about low order volume and suggest solutions.

Seamless coordination: Auto‑generate IM templates for dispute appeals, acting as a “digital lawyer”.

3.2 Passive Safety

1) Robust physical protection – The helmet meets national motorcycle helmet standards (GB 811‑2022 A3), exceeding the e‑bike standard. It absorbs impact, reduces head injury risk, and improves rider visibility with reflective accessories.

2) Fall‑monitoring – Sensors and algorithms detect collisions, trigger alerts, record audio/video, and initiate emergency calls. The workflow involves edge (helmet), side (mobile app), and cloud components:

Helmet detects abnormal motion via accelerometer and sends a suspicion event.

App fuses phone sensors to confirm the incident and records surrounding audio/video.

If the rider confirms a minor incident, the case is closed; otherwise, the app escalates, offering one‑click 120 dialing and AI‑driven verification.

Severe cases are pushed to platform safety staff for manual intervention.

Collected data (speed, location, sensor parameters) supports accident reconstruction, insurance assistance, and high‑risk road warnings.

04. Summary and Outlook

Meituan’s smart helmet has demonstrably improved delivery efficiency and safety by replacing phone interaction with voice AI and by providing real‑time safety monitoring. Riders report higher productivity and reduced distraction, especially during peak hours and complex traffic conditions.

In 2025, Meituan plans a next‑generation helmet that serves as a multimodal AI model gateway and data collection platform, offering richer environmental perception and further transforming the on‑demand delivery industry.