Singapore car rental business Lumens taps AI to boost productivity and lower costs
The app will eliminate human error and subjective judgment in the process of checking cars for damage
[SINGAPORE] Local car rental company Lumens faced a daunting operational challenge in documenting damage during handovers for its car rental business Lylo.
A car check took an average of 10 minutes, with staff manually documenting scratches and dents on paper forms. As the manual process is prone to human error and subjective judgment, the company hit upon the idea to tap generative artificial intelligence (Gen AI) to automate the process.
Lumens, an early mover in the private-hire vehicle industry, has been in business since 2014. With about 6,000 cars, it is now one of the largest fleet operators in Singapore.
It provides long-term rentals to private hire vehicle drivers, besides other services such as limousine services and short-term car rental.
In 2020, to boost its driver-partners’ incomes during Covid-19 when demand for ride-hailing services declined, Lumens set up Lylo, which counts among its clients corporate organisations, which can call on the drivers for their point-to-point transportation needs.
For example, after Singapore Expo realised that visitors faced difficulties in securing a ride when leaving the venue, Lylo worked with it to set up a streamlined process that enabled visitors to easily book a ride.
A NEWSLETTER FOR YOU
Friday, 8.30 am
SGSME
Get updates on Singapore's SME community, along with profiles, news and tips.
As Lumens grows and looks for areas to improve efficiency, it is also considering AI solutions for areas such as predictive maintenance and for personalisation of preferences for each customer.
“A good example at the intersection of operational efficiency and customer experience is car checks – it is critical because it carries financial implications for both Lumens and our customers,” Sun Zhibo, chief product officer at Lumens, told The Business Times.
The solution for the automated reporting of damage was developed last year as part of the Infocomm and Media Development Authority’s Gen AI x Digital Leaders initiative to help enterprises tap Gen AI with access to resources and expertise. While Lumens has an in-house technical team, it lacked the expertise to build it up on its own.
Instructions in an app will guide customers on how to capture photos from different angles, with AI models checking against the photos taken during collection to determine if any damage occurred. If damage is detected, the system will check against the rental insurance the customer has purchased to calculate the additional repair costs.
The system can also detect changes in the odometer and fuel gauge, verifying if customers have topped up the fuel tank to the same level when returning a car. Being integrated into the billing system, this makes the entire process seamless, with the data synchronised with Lumens’ fleet management system.
This reduces the amount of time needed to do manual car checks, which take about 10 minutes per car. The new system, with customers performing their own car checks, can quadruple productivity of the operational staff, allowing them to serve up to four customers simultaneously.
Sun said: “We estimate we will save more than S$300,000 annually, and this is just on the operational efficiency front.”
The new process will improve convenience, trust and transparency, which are key factors for customers. Lumens hopes that AI will aid in capturing a new revenue stream through this self-service process and attract a broader customer base in the growing digital mobility segment.
Developing this solution posed challenges, one of which was training the AI model to identify the right car model. This includes identifying the year in which it was manufactured, as the same car model across different years have different body shapes.
The AI models also have to learn the different dashboard layouts, which also can differ across different manufacturing dates for the same car model. This is further complicated as some dashboards have analogue gauges while others are digital.
“Additionally, lighting and weather conditions, and the photo angles will also significantly impact the assessment results. So that’s why the team actually spent a lot of effort in model training,” said Sun.
Spoofing detection measures are also needed to ensure the photos taken are of the same vehicle. Solving these challenges can be rewarding, said Sun, and pushes the AI solution to be more robust and reliable.
Lumens can tap the advantage of having over 10 years of car-check records in paper form along with photos of the damaged vehicles for training the AI model.
This solution is slated to be released towards the end of the year, and is expected to help with surges in demand during holiday periods.
Currently, the key focus is on enhancing the accuracy for the more challenging models, especially under various lighting conditions. Lumens is building an intuitive solution where the AI will guide the users through the car collection and car return process via a chatbox.
These measures are all part of plans to make the process of serving the customer better and grow the business.
Copyright SPH Media. All rights reserved.
