We measure how motorcycles are actually ridden, before a crash happens.
RideScore turns low-cost IMU and GPS data into rider safety scores, road risk maps, and before/after reports that show whether safety training and incentives actually work.
라이드스코어는 보급형 IMU 센서와 GPS로 이륜차 위험운전을 측정해 라이더 안전점수, 도로 위험지도, 교육 전후 효과 분석을 제공하는 이륜차 AI 안전 솔루션입니다.
Contact usThe problem
Delivery riding is one of the most dangerous jobs in Korea, yet nobody measures riding behavior before a crash. Commercial trucks and buses have digital tachographs (DTG) and risk-driving analytics. Two-wheelers do not, and four-wheel scoring misreads normal leaning turns as dangerous.
Baemin's delivery logistics subsidiary has had the most approved industrial-accident claims of any workplace in Korea (2022 to 2025).
Motorcycle traffic crashes per year in Korea, with about 500 deaths per year (2018 to 2022 average).
Personal-injury premium cut pushed by Korea's Financial Supervisory Service motorcycle insurance reform (Dec 2025), without behavior data to price risk.
Sources: Korea Workers' Compensation and Welfare Service data via National Assembly, Korea Road Traffic Authority, Financial Supervisory Service (2025.12).
What RideScore delivers
One riding-data platform for the three groups that pay for two-wheeler safety.
Delivery platforms
Personalized safety training based on each rider's own risk patterns, plus real-time risk alerts.
Public agencies
Quantified accident-prevention effects by comparing risky-riding rates before and after training.
Insurers
A two-wheeler-specific safety score to price risk and limit losses from mandated premium cuts.
Technology
Built on peer-reviewed research on delivery-rider safety.
Collect
Low-cost IMU + GPS units mounted anywhere on the bike (under the seat, on the delivery box), with smartphone sensing in parallel.
Calibrate & detect
Two-stage self-supervised calibration restores the vehicle frame (gravity alignment error 7.65° to 1.99°, about 1/5 the compute of a Kalman filter). Three-layer detection: data-quality rules, time-series anomaly detection, and motorcycle-dynamics checks (friction circle, sideslip). Controlled-test F1 0.73.
Score & report
Rider safety score (0 to 100, 5 grades), link-level road risk map, and intervention effect analysis.
- First-author paper in Accident Analysis & Prevention; co-author in Transportation Research Record
- PhD dissertation on two-wheeler riding safety, Hanyang University (2026)
- Founder is the inventor on two pending patent applications (sensor misalignment correction, hierarchical anomaly detection)
- Selected for Antler AI Residency, Seoul (2026)
Contact
Yeseo Gu, PhD, Founder
yeseo@ridescorelab.com · LinkedIn
Seoul, Republic of Korea