Smart sensing and mobility analysis for improved transportation efficiency

• Main Author: Cao, Chu
• Other Authors: Li Mo
• Format: Theses and Dissertations
• Language: English
• Published: 2019
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/106785; http://hdl.handle.net/10220/49682; https://doi.org/10.32657/10220/49682
• Institution: Nanyang Technological University

Sensors proliferate in human daily life and has changed the lifestyle of human beings. As the connection between human beings and digital devices, sensors not only bring new “ability” to users, such as localization, but also benefit users, such as activity monitoring. Therefore, one important research direction is to fully utilize the abilities of sensors built-in digital devices to provide further help for users. Many emerging applications leverage the sensors embedded in smart phones to provide service for users, e.g., indoor navigation, indoor/outdoor detection, visible light communication. All these applications rely on the powerful built-in sensors on smart phones. Analyzing the data generated by sensors during the daily usage of human beings can also reveal the knowledge hidden behind, e.g., tracking the daily activity to evaluate the health of users. Data collected from crowdsensing bring new opportunity to discover new knowledge that cannot be revealed using a small set of data. We observe that current navigation systems show digital maps rather than real world scene to users. Thus we present Amateur, an augmented reality based vehicle navigation system using commodity smart phones. Amateur reads the navigation information from a digital map, matches it into live road condition video captured by smart phone, and directly annotates the navigation instructions on the video stream. The Amateur design entails two major challenges, including the lane identification and the intersection inference so as to correctly annotate navigation instructions for lane-changing and intersection-turning. In this project, we propose a particle filter based design, assisted by inertial motion sensors and lane markers, to tolerate incomplete and even erroneous detection of road conditions. We further leverage traffic lights as land markers to estimate the position of each intersection to accurately annotate the navigation instructions. We develop a prototype system on Android mobile phones and test our system in a total number of more than 300 km travel distance on different taxi cabs in Singapore. The evaluation results suggest that our system can timely provide correct instructions to navigate drivers. Our system can identify lanes in 2s with 92.7% accuracy and detect traffic lights with 95.29% accuracy. Overall, the accuracy of the navigation signs placement is less than 105 pixels on the screen throughout the experiments. The feedback from 50 taxi drivers indicates that Amateur provides an improved experience compared to traditional navigation systems. When providing the navigation services to drivers, digital map plays a key role in navigation. The completeness of maps is significantly important. Most digital maps are designed for vehicles and miss a great number of walkways that can facilitate people’s daily mobility as pedestrians. Despite of such a fact, most existing map updating approaches only focus on the motorways. To fill this gap, we present VitalAlley, a walkway discovery and verification framework with mobility data from large scale crowdsensing. VitalAlley aims to identify the uncharted walkways from the big but noisy personal mobility data and incorporate these findings into existing incomplete road maps. The implementation of VitalAlley faces the major challenges due to the unstructured nature of the walkways themselves and the noise from crowdsensing data. VitalAlley leverages different aspects of individual mobility to model and estimate the walkable areas, based on which representative walkways that connect known road segments or points of interest are extracted. To verify the new-found walkways, we further propose image based auto-verification with the help of publicly accessible street image database from GSV. VitalAlley is implemented and evaluated with real world crowdsensing data from the Singapore National Science Experiment. As a result, 736 walkways (totaling 161 km in distance) are identified from the mobility dataset collected from 108,337 students in Singapore. We manually verify 224 walkways totaling 32.4km over a 9 km2 district through on-site inspection. The results suggest over 96% accuracy of VitalAlley in discovering the walkways.