Machine Learning Approach for Automated Detection of Irregular Walking Surfaces for Walkability Assessment with Wearable Sensor

Abstract

The walkability of a neighborhood impacts public health and leads to economic and environmental benefits. The condition of sidewalks is a significant indicator of a walkable neighborhood as it supports and encourages pedestrian travel and physical activity. However, common sidewalk assessment practices are subjective, inefficient, and ineffective. Current alternate methods for objective and automated assessment of sidewalk surfaces do not consider pedestrians’ physiological responses. We developed a novel classification framework for the detection of irregular walking surfaces that uses a machine learning approach to analyze gait parameters extracted from a single wearable accelerometer. We also identified the most suitable location for sensor placement. Experiments were conducted on 12 subjects walking on good and irregular walking surfaces with sensors attached at three different locations: right ankle, lower back, and back of the head. The most suitable location for sensor placement was at the ankle. Among the five classifiers trained with gait features from the ankle sensor, Support Vector Machine (SVM) was found to be the most effective model since it was the most robust to subject differences. The model’s performance was improved with post-processing. This demonstrates that the SVM model trained with accelerometer-based gait features can be used as an objective tool for the assessment of sidewalk walking surface conditions.

Deep-Learning-Based Approach for Automated Detection of Irregular Walking Surfaces for Walkability Assessment with Wearable Sensor

“…Kobayashi et al predicted different sidewalk surface types using random forest and acceleration data from smartphones stored in pedestrians' front pockets [25]. In our previous study [10], we proposed a traditional machine learning approach that analyzes hand-crafted accelerometer-based gait features extracted from a right-ankle sensor.…”

“…To address the gaps in existing works and to improve the prediction performance of the traditional machine-learning-based classification framework from our previous study [10], we introduced a novel deep-LSTM-based classification framework to automatically detect irregular walking surfaces with a single wearable sensor in this paper. In our previous study [10], we identified the right ankle as the ideal location for sensor placement for detecting irregular walking surfaces.…”

“…To address the gaps in existing works and to improve the prediction performance of the traditional machine-learning-based classification framework from our previous study [10], we introduced a novel deep-LSTM-based classification framework to automatically detect irregular walking surfaces with a single wearable sensor in this paper. In our previous study [10], we identified the right ankle as the ideal location for sensor placement for detecting irregular walking surfaces. Hence, in this paper, we used the acceleration data collected from a right-ankle sensor from our previous experiment and the traditional machine learning approach as the benchmark to validate the feasibility of our proposed deep LSTM framework.…”

“…To the best of our knowledge, none of the existing works explored incorporating gait analysis techniques with machine learning or deep learning approaches for sidewalk surface assessments. To bridge this gap, in our previous study [10], we identified the ideal body location for placing a sensor, which is at the right ankle. Then, we developed a traditional machine-learning-based classification framework for identifying irregular walking surfaces using gait features extracted from the right-ankle sensor to train machine learning models and demonstrated the effectiveness of the approach [10].…”

“…To bridge this gap, in our previous study [10], we identified the ideal body location for placing a sensor, which is at the right ankle. Then, we developed a traditional machine-learning-based classification framework for identifying irregular walking surfaces using gait features extracted from the right-ankle sensor to train machine learning models and demonstrated the effectiveness of the approach [10]. Nonetheless, the traditional machine-learning-based method requires manual feature extraction and selection, which can be time-consuming and labor-intensive.…”