How Yulu Manages the Traffic Menace it Creates in Bengaluru

In Bengaluru, when using a quick commerce service, one must have witnessed delivery representatives riding small blue electric scooters to deliver goods. Sometimes, these scooters can be spotted at parking spots on the sidewalk or lying around near the curb, which often creates a menace for the traffic.

These electric scooters were built by Yulu, a smart shared mobility startup founded in 2017 by Amit Gupta, Naveen Dachuri, RK Misra and Anuj Tewari, to solve the simple problem of short-distance mobility. Dachuri, the CTO of the company, told AIM that he is aware of the inconvenience that these scooters sometimes create in the city and assured that the company handles it as much as possible from its end.

While Dachuri delved a little into how the company plans to integrate chatbots in Indian languages on scooters, much of the plans are still in the ideation stage.

Real-Time Predictive Models

“If you have to run smart mobility, the most important thing is…to have some kind of intelligence built in the vehicle so that the manual interferences can stop,” Dachuri said. According to him, this can only be achieved if there is a device built inside the vehicle which can give information in real time about the whereabouts and the condition of the vehicle.

Off-the-shelf products did not work for Yulu, as the changes needed for their business took months. To address this challenge, the company developed its proprietary Internet of Things (IoT) system, Yulu Connect. 

This system gathers vehicle data and performs edge computing, which enables real-time decision-making. It can do this both while the ride is on and the vehicle is parked. This ensured no delay and no compromise of the customer experience.

Real-time data is primarily used to enhance customer experience, such as guiding users to the nearest battery-swapping station based on availability. MySQL is used for real-time data storage, while telemetry and sensor data flow into TiDB, a distributed SQL database. 

Additionally, streaming engines process live data to enable actions such as dispatching field executives for maintenance and directing users to optimal parking spots.

Apart from using the data in real time, which is purely based on the current state of the vehicle, Yulu also uses the data to build predictive models to increase the uptime of the vehicle and improve the efficiency. This also helps in real-time prediction based on historical information. 

“We match these two pieces of information and tell users to go to a certain location so that your probability of getting a battery or a vehicle will be higher,” Dachuri explained.

Reducing Repair and Maintenance

The predictive models made by Yulu are beneficial for scheduling vehicle repairs and maintenance. The problem with shared mobility is that when original equipment manufacturers (OEMs) provide the vehicles, the learnings about the possible wear and tear are based on personal vehicles and not shared usage, where multiple people are using the same vehicle. 

This is where Yulu’s data comes into the picture. Dachuri and his team have collected data over the years about how the different types of braking, throttling, riding over potholes, and even sitting posture affect the long-term condition of the vehicle. “Everyone wants to ride in a smooth fashion. But unfortunately, the conditions make you take those routes. And that is how I at least see with respect to Bengaluru roads.”

Addressing the issue of parking and vehicle mismanagement, Dachuri said that his team built a geofencing framework which encourages users to park at designated stations. However, geolocation accuracy remains a challenge due to factors like satellite visibility and urban infrastructure interference. 

To counter this, Yulu combines IoT device data with users’ smartphone GPS to optimise parking suggestions. However, this also results in significant errors and drainage of the vehicle’s battery when the location is optimised. 

“We suggest operators to use the most optimal location accuracy, but oftentimes they end up choosing something else,” Dachuri said. He added that despite technological solutions, user education remains vital in ensuring compliance with parking guidelines.

The Safety Issue

By the time all these things are addressed, a lot of menace is already created. Yulu often also suffers from regulatory problems from the government, such as the seizure of vehicles or hefty fines because of the created menace by the gig workers, including driving on the wrong side, among several other violations.

To address a lot of these road safety issues, Yulu has developed a system to detect wrong-way driving. Initially relying on Google APIs, the company found them insufficient due to frequent road direction changes. Consequently, they built an in-house solution using simple mathematical models to detect wrong-way movements and issue warnings and penalties to violators. 

Dachuri, however, said that many of these gig workers do it to deliver multiple orders from dark stores at the same time, which forces them to travel on the wrong side. “People opt for more deliveries than safety.” Dachuri added that the team tries to advise them, but when it comes to the tech stack, they have done much of what could be done.

Yulu also employs strategic field operations to maintain order. “During low-traffic hours, field executives reposition vehicles and ensure cleanliness. Signages and designated parking zones are regularly monitored, though evolving city infrastructure poses challenges in keeping them updated,” Dachuri explained.

Yulu also offers ISA-certified helmets for drivers, which they can buy at a low price to ensure their safety. The only problem that still remains, according to Dachuri, is the education and awareness of Yulu users; the tech part is now mostly fixed.