Continental_Sensor System

Team members

Wong Zhi Cong (EPD), Lui Yan Le (EPD), Jerome Heng Hao Xiang (ISTD), Yai Yan Lin (ISTD), Gwee Yong Ta (ISTD), Ong Kah Yuan Joel (ISTD), Tan Joon Kang (ISTD)

Instructors:

Ye Ai, Cyrille Pierre Joseph Jegourel

Writing Instructors:

Rashmi Kumar

Teaching Assistant:

Congjian Lin

Problem: Urban Inspection in Singapore

Current urban inspection methods in Singapore are uncoordinated, labour intensive and costly .

Lack of Coordination

Division of responsibility between government agencies results in difficulty in coordinating inspections.

Labour Shortages

Private contractors face rising costs of labor, making frequent inspections a costly procedure.

Uninformed Decisions

Manpower allocation decisions are mostly based on intuition and experience. There is also little data available to establish trends.

Solution: Street Scanner

Street Scanner is a sensor boom specifically made for pavement inspection and maintenance. By using a combination of sensors, cameras and artificial intelligence, it automates the tedious inspection process and seamlessly integrates into current practices.

Street Scanner is attached to Corriere, Continental's last mile delivery robot. Besides fully automated inspection, it also provides a web and mobile interface that can be accessed by different stakeholders to view and interact with inspection data.

Patrolling

Corriere is dispatched to a certain sector, where it patrols and captures images with its onboard cameras.

Inspection

As it patrols, Corriere uploads images to the cloud, where pavement defects and litter are detected and flagged

Management

Using our web dashboard, Managers can see where these detections happen, and can assign workers to rectify the issues

Rectification

Workers will be notified via mobile app when assigned to a task by managers. They can use the app to upload photos of rectifications for approval.

Deployment and Data Collection

Corriere has powerful, high-resolution cameras that can capture quality images of pavement and litter and upload them to the cloud.

Coupled with its long battery life and quick charge time, it can collect an abundance of high-quality inspection images in a single journey.

Litter and Crack Detection

Our AI model uses state-of-the-art models such as YOLOv5. It is trained on a total of 8,000 images from datasets like TACO (Trash Annotations in Context) and various pavement datasets, allowing it to detect pavement cracks and common litter with high accuracy.

Being deployed on the cloud, it offloads computationally intensive tasks from Corriere's onboard computer, and can also easily be swapped out with other AI models for various detection purposes.

Fully Equipped Dashboard

Street Scanner is also supported by a fully equipped web dashboard that allows managers to view inspection data.

The dashboard is able to display trends and statistics of past inspection data, giving managers better insight on the general state of the area over time. It also shows tasks assigned to workers so that they can track and review their progress.

Realtime Updates

With an informative map view, managers can easily see realtime inspection updates as Corriere embarks on its journey, all from a single screen.

Contractors can immediately take note of areas that need attention and allows them to make swift and informed decisions on when and where to deploy workers.