Micro-Mobility Safety

Cycling must be safe, and perceived as such, if micro-mobility trips by all populations are to increase, and the benefits in traffic decongestion and carbon emission cut are to be realized. WHO estimates that 40,000 cyclists died in road accident in 2016. There is an urgent need to address this problem by low-cost and robust safeguard systems.

We developed riding safegard with regard to two categories: (a) Distracted behavior recognition (b) Riding Maneuver Prediction.

HEADSENSE

We present HeadSense (Han et al., 2023), a helmet-based system that leverages the inertial motion unit (IMU) to recognize 4 distracted behaviors: checking handlebar-mounted devices, using smartphones, attracting to the roadside element, and abreast riding. The feasibility lies as followed: - The head dynamic is a strong indicator of the rider's eyesight / attention. And the helmet, as an essential bicycle accessory, follows the same movement pattern as rider's head. - The rider’s head shows unique motion patterns during distracted riding behaviors. These patterns consist of a series of abnormal head movements. Evaluation results show HeadSense can segment visual search into episodes with an accuracy of up to 86.14%. Additionally, from sequences of episodes, it can effectively detect distracted riding behaviors at an average precision of 85.04%.

HEADMON

Detection of ongoing maneuver may be less effective in accident prevention. In this work (Han et al., 2023) (Han et al., 2024), we take a step further to explore the feasibility of using riders’ head dynamics or handlebar motion to predict their riding maneuvers with two key observations: Rider needs to observe the traffic situation in advance based on their riding maneuver intentions. For different maneuver intentions, the rider's head dynamics (such as turning left and right) are also different. We constructed an Attention-based network to solve the prediction problem. The precision of riding maneuver prediction is at least 0.80 under 4 seconds time gap. We also finds that the accuracy would be improved with longer detection window size. The results indicate a novel start on improving micro-mobility safety.

DOUBLECHECK

DoubleCheck (Dong et al., 2022) is a method that utilizes a handlebar-mounted smartphone to detect single-handed cycling and followed distracting secondary tasks. The work was established on the premise that single-handed cycling undermines the stability of handlebar during cycling. Preliminary data shows that For both acceleration and angular speed: - The signal has denser power over the frequency band during single-handed cycling. - The signals contain periodic components Accordingly, we adopt Autoregressive Model, featuring robust performance in extracting features of periodic time-series signal. Experiment with 22 participants on asphalt and pavement demonstrated that DoubleCheck achieves an F1-score of 0.96 for hand detection and 0.69 for distraction recognition.

References

2024

RideGuard: Micro-Mobility Steering Maneuver Prediction with Smartphones

Zengyi Han , Xuefu Dong, Liqiang Xu , and 4 more authors

In 2024 IEEE 44th International Conference on Distributed Computing Systems (ICDCS) , 2024

Acceptance Rate 21.9%

2023

HeadSense: Visual Search Monitoring and Distracted Behavior Detection for Bicycle Riders

Zengyi Han , Xuefu Dong, Yuuki Nishiyama , and 1 more author

In 2023 IEEE 24th International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM) , 2023

Acceptance Rate 28.2%

Abs PDF

Distracted riding behavior is one of the main causes of bicycle-related traffic accidents, resulting in a large number of casualties and economic losses every year. There is an urgent need to address this problem by accurately detecting distracted riding behaviors. Inspired by the observation that distracted riding behaviors induce unique head motion features that respond to the rider’s attention, we present the HeadSense, a helmet-based system that not only monitors the visual search episode of the rider but also detects distracted riding behaviors. Specifically, HeadSense leverages the inertial motion unit (IMU) to recognize distracted behaviors such as using smartphones, attracting to the roadside element, and abreast riding. We designed, implemented, and evaluated HeadSense through extensive experiments. We conducted experiments with 19 participants inside the university’s campus. The experimental results show that HeadSense can achieve an overall accuracy of 86.14% while monitoring visual search episodes. Moreover, HeadSense can detect the occurrence of distracted riding behaviors with an average precision of up to 85.04%.
HeadMon: Head Dynamics Enabled Riding Maneuver Prediction

Zengyi Han , Liqiang Xu , Xuefu Dong, and 2 more authors

In 2023 IEEE International Conference on Pervasive Computing and Communications (PerCom) , 2023

Acceptance Rate 17%

Abs PDF

Although micro-mobility brings convenience to modern cities, they also cause various social problems, such as traffic accidents, casualties, and substantial economic losses. Wearing protective equipment has become the primary recommendation for safe riding. However, passive protection cannot prevent the occurrence of accidents. Thus, timely predicting the rider’s maneuver is essential for active protection and providing more time to avoid potential accidents from happening. Through the qualitative study, we argue that we can use the rider’s head dynamic as an information source to predict the rider’s following maneuvers. We accordingly present HeadMon, a riding maneuver prediction system for safe riding. HeadMon utilizes the head dynamics of a rider by installing an inertial measurement unit on the helmet. It uses the extracted head dynamics features as the input of the deep learning architecture to achieve prediction. We implemented the HeadMon prototype on Android smartphone as a proof of concept. Through comprehensive experiments with 20 participants, the result demonstrates the excellent performance of HeadMon: not only could it achieve an overall precision of at least 85% for maneuver prediction under a 4s prediction time gap, but it also could keep a high accuracy under a low sampling rate. The low-cost feature of HeadMon allows it to be readily deployable and towards more safety riding.

2022

DoubleCheck: Detecting Single-Hand Cycling with Inertial Measurement Unit of Smartphone

Xuefu Dong, Zengyi Han , Yuuki Nishiyama , and 1 more author

In 2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) , 2022

Abs PDF

Riding bikes with only one hand on the handlebar can severely undermine the steering capability of riders and risk road safety. In this study, we propose a first detection framework for monitoring single-hand cycling on bicycle travel, called DoubleCheck. It is based on the premise that riders adapt their body movement during single-hand cycling, which is distinguishable to the sensors even amid noise from the exasperate road surface. The system can detect handlebar-holding under different road conditions using motion signals from a built-in inertial measurement unit (IMU) in a handlebar-mounted smartphone. We implemented the system and invited 10 participants for our evaluation experiment. Our results show that DoubleCheck achieved an F1-score of 0.94 for hand detection, proving its efficacy for real-life implementation to improve road safety.