Enhancing micromobility safety: an advanced sensing system and a data-driven approach for hazards detection

Road safety remains a critical concern for vulnerable road users (VRUs) including pedestrians and employees of courier companies. Additionally, safety of micromobility devices represents one of the most significant factors that influence the adoption of this mode of transportation in urban environments. However, micromobility users such as cyclists and motorcyclists may face various types of risks and hazards that may cause injury or result in fatality. These hazards include road accidents, dangerous intersections, conflicts with other road users and poor road infrastructure. To enhance the safety of micromobility and VRUs, UoW in the context of SOTERIA project introduces an innovative sensor kit which can be mounted on microvehicles to collect micromobility-related data, and proposes a data-driven approach to detect micromobility hazards in real time. This approach may enhance the road safety and improve the situational awareness of VRUs through identifying risky areas and dangerous intersections.

SOTERIA Sensor Kit

Sensing systems may be mounted on microvehicles such as bicycles to collect safety-related data including experienced vibration events, proximity to the surrounding objects, speed changes, riding styles and behaviours. SOTERIA Sensor Kit which represents an advanced technology has been designed and implemented by integrating several sensors and devices including inertial measurement unit (IMU), global positioning system (GPS), grip force sensor and depth camera. The Sensor Kit is mounted on a bicycle to capture accurate and reliable real time data on bicycle motion, road conditions, user behaviour, traffic pattern and environmental conditions. Therefore, this collected data can be used to measure multiple aspects of microvehicle handling (i.e. steering, accelerating and braking) and support a range of safety services and solutions. The sensor kit and advanced operational analytics implies considerable and effective advantages such monitoring user behaviours, detecting road hazards and identifying traffic conditions. Additionally, a web interface has been developed for the Sensor Kit so the real time data collected by the different sensors and devices can be accessed and the related signals can be monitored.

Micromobility hazards detection

Hazards detection and mapping is one of the VRUs and micromobility situational awareness safety services developed in SOTERIA project. It represents a data-driven service which uses the data collected by the Sensor Kit to generate a real time micromobility hazards map. The hazards map shows various types and severities of hazards that may affect the wellbeing and comfort of micromobility users, destabilise the microvehicle or accumulatively lead to a potential crash. Detected hazards mainly include areas with specific infrastructure deficiencies such as potholes and rough-surface paths, obstacles, and risky intersections. Therefore, the hazards map supports micromobility users in their day-to-day traveling and in minimising risks lead to accidents by enabling these users to avoid dangerous locations and potential accidents. The hazards detection approach involves data preparation and processing, features extraction, data labelling, features reduction, data re-labelling and machine learning models (ML) training. Effectively, videos captured by depth camera are analysed and used to label the data by the hazard type while ML clustering is used to re-label the data by the hazard severity.

The piece has been authored by the University of Wolverhampton.