Creating a smart system for vehicle fleets
1
CME Professor Jane Lin is investigating a way to monitor a variety of vehicle fleets accurately to ensure safety and good service.
The research is supported by a grant from the US Department of Transportation titled “SBIR: FleetLynq: AI-driven Decision Support Solution for Transit Fleet.” Lin is partnering with the company Flexlynqs, which was founded by UIC graduate Santosh Mishra, who earned his master’s in transportation from UIC in 2004.
Flexlyng’s mission is to empower public sector agencies to plan, design, and deploy mobility solutions with the latest advancements in technology, keeping human factors in mind so that anyone can go anywhere with sustainable modes prioritized to reduce reliance on single-occupancy vehicles.
Most transit agencies keep approximately 10-15% extra vehicles as backups, and more than 25% of their buses are close to the end of their remaining useful life. The fleet maintenance relies heavily on historical failure data and manual inspections, limiting the ability to detect issues before they escalate. Due to reactive repairs, industrial experts estimate that downtime costs $448 – $760 per vehicle daily.
“As vehicles age, the telemetry data from vehicle components gets noisy as vehicles age and sensors drift, leading to incorrect diagnoses,” said Lin, director of the SusTrans Lab at UIC.” Sensor drift in aging vehicles leads to a 20-30% surge in false-positive diagnostic alerts, prompting unnecessary servicing.”
As agencies switch to cleaner vehicles — like compressed natural gas, battery-electric, or hydrogen fuel cell buses — and start using more automation, solving problems becomes even harder. These newer vehicles have lots of electronics and can generate millions of pieces of data every day. When issues are misdiagnosed, buses can break down unexpectedly, causing service delays and inconvenience for riders.
Because of this, it’s tough for agencies to manage their fleets efficiently while still providing reliable service. The shortage of skilled technicians makes it even harder.
“Our goal is to create a smart system that uses artificial intelligence and machine learning to help diagnose vehicle issues. The system will include a self-learning knowledge base that improves over time,” Lin said. “While it will use Edge AI devices to process data, the system — called FleetLynq — will be flexible and work with different types of hardware and cloud setups. It will choose the best way to analyze data depending on each fleet’s size, needs, and budget.”
The proposed predictive maintenance system will help reduce maintenance and operating costs of transit agencies and improve the safety and efficiency of the transit operation with fewer breakdowns and failures of the fleet, providing a safe and reliable transit service, which is a vital part of a good economy.