Abstract:Pedestrian heading tracking enables applications in pedestrian navigation, traffic safety, and accessibility. Previous works, using inertial sensor fusion or machine learning, are limited in that they assume the phone is fixed in specific orientations, hindering their generalizability. We propose a new heading tracking algorithm, the Orientation-Heading Alignment (OHA), which leverages a key insight: people tend to carry smartphones in certain ways due to habits, such as swinging them while walking. For each smartphone attitude during this motion, OHA maps the smartphone orientation to the pedestrian heading and learns such mappings efficiently from coarse headings and smartphone orientations. To anchor our algorithm in a practical scenario, we apply OHA to a challenging task: predicting when pedestrians are about to cross the road to improve road user safety. In particular, using 755 hours of walking data collected since 2020 from 60 individuals, we develop a lightweight model that operates in real-time on commodity devices to predict road crossings. Our evaluation shows that OHA achieves 3.4 times smaller heading errors across nine scenarios than existing methods. Furthermore, OHA enables the early and accurate detection of pedestrian crossing behavior, issuing crossing alerts 0.35 seconds, on average, before pedestrians enter the road range.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: **How to achieve reliable pedestrian orientation tracking in pedestrian crossing prediction, especially when using ordinary devices (such as smart phones), how to deal with the challenges brought by the changeable phone postures**.
### Specific problem description:
1. **Limitations of existing methods**:
- Existing pedestrian orientation tracking methods mainly rely on inertial sensor fusion or machine - learning techniques, but these methods usually assume that the phone is fixed in a specific posture, which limits their universality.
- When the user changes the posture of the phone during walking (for example, swinging the phone, putting it in the pocket or bag), existing methods are difficult to accurately track the pedestrian's orientation, resulting in increased errors.
2. **Specific application scenarios**:
- Pedestrian crossing prediction is an important application scenario, which can improve the safety of road users. However, due to the complex and diverse behaviors of pedestrians crossing the road and the different phone - using habits of different users, existing orientation tracking methods perform poorly in this task.
### Proposed solutions:
To address the above problems, the paper proposes a new orientation tracking algorithm - **Orientation - Heading Alignment (OHA)**. The main innovation points of this algorithm include:
1. **Utilizing user habits**:
- OHA takes advantage of people's habits of carrying smart phones (such as swinging the phone, putting it in the pocket, etc.), and improves the accuracy of orientation tracking by learning these habitual postures.
- For each smart phone posture, OHA maps the direction of the phone to the pedestrian's orientation and efficiently learns this mapping relationship from the rough orientation and phone posture.
2. **Real - time prediction of pedestrian crossing behaviors**:
- The author applies OHA to a challenging task: predicting when pedestrians are about to cross the road. By analyzing 755 hours of walking data (from 60 participants), a lightweight model has been developed, which can achieve real - time prediction on ordinary devices.
- The experimental results show that the orientation error of OHA in nine different scenarios is 3.4 times smaller than that of existing methods, and it can send a crossing - the - road alarm 0.35 seconds in advance on average before pedestrians enter the road area.
### Summary:
This paper aims to solve the problems of reliability and universality of pedestrian orientation tracking in practical applications, especially for the performance of ordinary devices (such as smart phones) in multiple postures. By proposing the OHA algorithm, the author not only improves the accuracy of orientation tracking, but also successfully applies it to pedestrian crossing prediction, significantly improving road safety.