RobotCycle: Assessing Cycling Safety in Urban Environments

TL;DR

RobotCycle tackles urban cycling safety by collecting ego-centric, multimodal data through a wearable backpack and analyzing how road infrastructure and traffic interactions shape cyclist behaviour. It leverages AV-inspired data collection and HD mapping to build trajectory models and a safety score that reflects infrastructure quality and traffic dynamics. The work introduces a novel wearable platform, an annotated Oxford and Culham dataset, and a benchmarking framework for cyclist safety and cyclability with the aim of informing infrastructure design. The intended impact is practical: enabling city planners and policymakers to identify hotspots and test cyclist-friendly interventions for safer, more sustainable urban mobility.

Abstract

This paper introduces RobotCycle, a novel ongoing project that leverages Autonomous Vehicle (AV) research to investigate how road infrastructure influences cyclist behaviour and safety during real-world journeys. The project's requirements were defined in collaboration with key stakeholders, including city planners, cyclists, and policymakers, informing the design of risk and safety metrics and the data collection criteria. We propose a data-driven approach relying on a novel, rich dataset of diverse traffic scenes and scenarios captured using a custom-designed wearable sensing unit. By analysing road-user trajectories, we identify normal path deviations indicating potential risks or hazardous interactions related to infrastructure elements in the environment. Our analysis correlates driving profiles and trajectory patterns with local road segments, driving conditions, and road-user interactions to predict traffic behaviours and identify critical scenarios. Moreover, by leveraging advancements in AV research, the project generates detailed 3D High-Definition Maps (HD Maps), traffic flow patterns, and trajectory models to provide a comprehensive assessment and analysis of the behaviour of all traffic agents. These data can then inform the design of cyclist-friendly road infrastructure, ultimately enhancing road safety and cyclability. The project provides valuable insights for enhancing cyclist protection and advancing sustainable urban mobility.

Figures (8)

• Figure 1: Data-collection platform design (a) and deployment (b). The sensors and computing unit have been integrated into a backpack, allowing application flexibility without compromising the specifications and FOVs of the sensors.
• Figure 2: Various design iterations of the RobotCycle backpack. From the left, the first iteration (a) allowed for sensors' displacement to optimise their position and orientation. The second iteration (b) consolidated the sensors' positioning, computing requirements, and networking, allowing extensive field testing. The final iteration (c), finalised the sensing configuration and design, maximising sensors' FOV and optimising overall weight distribution.
• Figure 3: Detail of the articulated systems that enabled extensive experimentation with various sensors' positioning to optimise the sensing setup.
• Figure 4: (Left) CyclOSMbicycle oriented map rendered from OpenStreetMap (OSM) OpenStreetMap. (Right) Visualisation of places of interest on Google Maps.
• Figure 5: Sensor samples from the second iteration of the backpack in \ref{['fig:iterations']}b. Raw 3D scans of the side lidars are visualised in (a), (b), and combined in (c); undistorted image data from the front and rear stereo cameras are shown in (d) and (f); raw images from the monocular side cameras are seen in (e) and (g). The side cameras are synchronised with the lidars through the Pulse Per Second (PPS) signal from the GPS and a custom signal repeater; GPS data can be seen in \ref{['fig:hd_map']}.