https://video.osgeo.org/videos/watch/e53bb7c0-6acc-45f8-aa43-ce2273c24e6d Context and Problem StatementUrban mobility is shifting toward sustainability, with cycling as a primary pillar. However, a significant barrier remains: the perceived danger in traffic. Traditional city planning relies on official crash reports, which ignore "near-miss" incidents and environmental factors that shape a cyclist's experience. Current technological solutions often face hurdles such as high data overhead, privacy risks (due to video recording), or technical complexity that prevents citizen participation.2. The Solution: senseBox System DesignTo bridge this gap, the authors developed an innovative, open-source sensor system based on the senseBox MCU. Designed for ease of assembly (solder-free), the device is mounted to the bicycle's seat post and includes:Environmental Sensors: Temperature, humidity, and particulate matter ($SPS30$).Safety/Motion Sensors: Accelerometer ($MPU6050$) and Time-of-Flight ($ToF$) ranging ($VL53L8CX$).Connectivity: BLE communication with a smartphone app to merge sensor data with geolocation.Privacy by Design: Users can set "privacy zones" and control data uploads to the openSenseMap platform.3. Edge AI: Machine Learning on the BikeA core innovation is the integration of TensorFlow Lite for on-device processing. This "Edge AI" approach minimizes bandwidth and power consumption while protecting privacy. Two main ML models are deployed:Overtaking Detection: Using shallow neural networks and the $8 \times 8$ multizone ToF sensor to identify dangerously close passing maneuvers by vehicles.Road Quality Classification: Using accelerometer data to classify surface types and detect roughness, using OpenStreetMap data as a ground-truth reference.4. Citizen Science and WorkshopsThe project emphasizes a participatory approach to bridge the gap between scientific research and community needs.São Paulo (Brazil): 20 participants built and mounted their own sensors, collecting data in a complex urban environment.Münster (Germany): Follow-up workshops allow for a comparative analysis of contrasting cycling infrastructures and perceptions.User Studies: Participants evaluate the system’s usability and data trust, comparing the AI’s recorded incidents against their personal feelings of safety during the ride.5. Conclusion and Future OutlookThe system provides a high-quality, spatial dataset that reflects the authentic experiences of cyclists. By empowering citizens to act as data collectors, the project offers actionable insights for traffic and transport planning. Future work involves developing an open-source recommender system for city officials, guiding data-driven improvements to infrastructure and overall road safety. Sun, 31 May 2026 09:47:23 GMT https://validator.w3.org/feed/docs/rss2.html PeerTube - https://video.osgeo.org https://video.osgeo.org/lazy-static/avatars/de841c88-3a57-45ac-9eff-a00237356bfc.png https://video.osgeo.org/videos/watch/e53bb7c0-6acc-45f8-aa43-ce2273c24e6d All rights reserved, unless otherwise specified in the terms specified at https://video.osgeo.org/about and potential licenses granted by each content's rightholder.