Highways and roads are a crucial part of every country’s public infrastructure. Their maintenance currently faces three main challenges. 1) daily road maintenance operations and technical status assessments of highways rely on tacit human experience. 2) inspection and maintenance workers are using portable terminal tools with low levels of intelligence. 3) on-site operations and back-office supervision and management of maintenance operations are inefficient due to slow turnover rates.
According to the Ministry of Transport, China’s public highways and roads system’s nationwide mileage was 5,198,100 kilometers as of the end of 2020, which is an increase of 185,600 kilometers year-over-year. The mileage considered to be public roads and requiring public sector maintenance is 5,144,400 kilometers, which accounts for about 99% of the total road mileage in the country. Hence, the public sector bears major responsibility for maintaining and expanding infrastructure.
The road surface inspection and maintenance management system are the current mainstream information solution, but heavily relies on maintenance crews and doesn’t fully exploit the value of data collected.
SenseTraffic ICV Maint is SenseTime’s dynamic AI solution with software and hardware components that supports road maintenance operations by offering comprehensive detection capacity for road owners, road operators, and third-party road maintenance companies with an end-to-end cloud architecture.
The data for the roads is collected by a vehicle mounted with edge computing platform. The collected data is transmitted by 4G and 5G networks through SenseTime’s SenseFoundry for processing, but not storage, to (1) produce API to enable the road maintenance defects detection capability, (2) enable the creation of an effective road defects management system, and (3) reliably classify defects detected through data-driven analysis and machine learning.
The analysis can yield a comprehensive typology of road surface defects, including surfaces that are cracked, filled with potholes, have lost structural integrity, have lower layers exposed, have rutting components, or undergoing repair. This efficient solution allows public sector decision-makers to deploy resources quickly to repair highways and roads in a timely manner for public safety and functionality for public good.
What Makes Your Project Innovative?
SenseTime was the first mover in this field, which gave the firm a strong market advantage. The firm’s extensive experience provided a solid foundation for the system to be robust and efficacious.
The current solutions, including human inspection patrols and human-operated inspection vehicle patrols, are more costly compared to the automated defect detection vehicle enabled by SenseTraffic ICV Maint and requires four to eight times the amount of time to complete the same task with 95% accuracy. The use of the SenseTraffic ICV Maint vehicle enable more efficient coordination between the inspectors and maintenance crews by providing timely information on defects found during road surveys. The SenseTraffic ICV Maint solution provides highly standardized results for review within one hour of scanning a section of road while human inspectors need four or more hours to report their findings.
What is the current status of your innovation?
The flexibility of SenseTraffic ICV Maint allows users to operate it as either a platform as a service (PaaS) or software as a service (SaaS) depending on the preference of the client. This solution’s all-in-one platform combined with edge computing supported by AI computing center ensures high efficiency and efficacy for clients to complete road surveys.
The solution has been piloted with success for at least two major highways near major cities in China. At least three firms have adopted SenseTraffic ICV Maint as a solution for highway and road maintenance.
Collaborations & Partnerships
In developing Sense Traffic ICV Maint, Sensetime has been working closely with public sector partners including local road maintenance agencies, state-owned infrastructure companies, and other technology firms in the private sector.
Users, Stakeholders & Beneficiaries
Road and highway maintenance agencies, smart infrastructure technology firm, state and local governments are the main users of the innovation.
The local and national governments and the public that uses the highways and roads are the stakeholders.
The main beneficiaries are the general public that uses highways and roads that are supported by SenseTime’s AI solution for road maintenance.
Results, Outcomes & Impacts
The performance of the SenseTraffic ICV Maint solution was demonstrated and validated at the IEEE’s Global Road Damage Detection Challenge 2020 during which this solution produced the best result in two competitions with 136 rival teams.
In two use cases involving surveys on major highways, SenseTraffic ICV Maint has shown strong performance by successfully detecting nine types of defects and transmitting corresponding station numbers and details on defects in real time to enable experts to review detected defects remotely with
high efficiency. Furthermore, these cases also showed that the solution can detect surface defects under rainy road conditions. Manual confirmation of the AI solution’s analysis showed that its accuracy was higher than 83.7%.
Challenges and Failures
The environmental conditions for each region where SenseTraffic ICV Maint is deployed are unique and requires time to train the model to optimally support road maintenance for that location.
Conditions for Success
Availability of rich data, faster connectivity, digital capacity, and cross-team cooperation are some of the main factors for the successful deployment of SenseTraffic ICV Maint.
The accumulation of data, tacit knowledge, and experience from the projects under way will serve to improve future deployments of SenseTraffic ICV Maint for scaling. Each project’s gains serve to improve the solution’s general efficacy and optimized uses for future projects to better meet new and current client needs.
- Implementation - making the innovation happen
25 November 2021