ZaiBike is a smart bicycle sharing system where users are able to rent a bicycle from their current location and return it at their destination. It differs from conventional station-based systems by allowing the user to park the bicycle anywhere. The parking system then evolves dynamically over time to optimise between user convenience, operational cost, and orderliness through IoT and data analytic applications.

My personal role in this startup venture is to lead the building of the business, product, and team. On the technical side, I played a part in developing the frontend mobile application and backend system, including the data analytics.

ZaiBike’s vision is to augment the Singapore transport landscape through data analytics by implementing a smart bicycle sharing system.

The bicycle is the most efficient human-powered form of transport. Source: Image credit:

Why the bicycle, though? The bicycle is a fantastic first mile last mile commuting device which allows the rider to get from his current location directly to his destination. In many cities around the world, and especially so in the urban jungle of Singapore, our homes and amenities are so well connected that you would only need a half an hour’s walk to satisfy all your requirements to live. However, a half an hour commute is already considered a lifetime to many urban dwellers, much less at a walking pace. Cars, buses and trains simply are not sustainable for these short distance commutes as population density grows, and thus, we welcome the bicycle. They help bridge this gap and expand your horizons, all without breaking much more of a sweat.

The sharing economy is also the most efficient way a resource can be utilised. Personally owned bicycles are often seen rotting at parking bays at HDBs and MRTs, and simply a hassle to own and maintain at times. Shared bicycles allow riders the convenience to pick up a bike wherever they are, and drop it at wherever they want to, with no burden from the sense of ownership. The combination of the two allows us to build an efficient mode of transportation all while efficiently utilising the available resources!

As an operator, the data that you are able to gain from these commuting trips are also invaluable. It allows us to build models of where people are coming from, where they are going to, and the route they are taking. Urban infrastructures can then be built around these findings to optimise travel time and travel experience!

With such a fantastic opportunity to make the world a better place, I embarked on this journey in 2015.

A lovely sunset view at East Coast Park

Its origins were far humbler, though. It started off as a pet project meant to enhance the quality of life of SUTD students upon moving to the East Coast Campus. With great food and parks in the East side, the prospect of cycling out for sumptuous meals and romantic walks along East Coast Park was tantalising. It also represented a great opportunity for us engineers to come together to create something in the real world during our university years too!

SUTD provided that great opportunity to help us get started, with the SUTD-MIT Internal Design Centre (IDC). IDC had a grant where innovative projects are funded to bring ideas into the real world, and that was where we secured the financial means to carry out the project.

We had to think of a way to innovate the bicycle sharing landscape, though. What is the use of copying another country’s bicycle sharing system and remaking it in our home? Our idea had to be innovative: it had to maximise the utility of these systems yet ward off its pitfalls.

After much research, deliberation, and riding around the Hangzhou and Boston bicycle sharing services, our eventual idea was to adopt a free-floating model, as opposed to the station-based models, built around the smart bicycle.

An example of the station-based bicycle sharing system: Hangzhou public bike share scheme. Image credit:

Our decision was based around 3 factors: cost, land ownership rights, and user convenience.


One of the main reasons for the unsustainability of many bicycle sharing systems in the world is the cost of building an electrically driven hub. New systems will find it hard to expand, while there will be no room for error in the placement of the docks. These costs will either have to be passed to the sponsors (government or likewise), or to users. Land space is also a premium in urban cities, often competing against other infrastructure, sometimes incurring the wrath of the public.

Land Ownership Rights

From our observations of international systems, the bicycle sharing systems were not present in privately owned land, such as in business parks and campuses. Singapore’s land ownership rights are equally, if not more, complex, with different paths, roads, and districts owned by different government agencies and private developers. This diminishes the objective of a bicycle sharing system that aims to provide riders from their location directly to their destination, resulting in a loss of market penetration.

User Convenience

As users, we often found upon reaching our destination on a shareable bicycle that the station was full, or we still needed to walk a considerable distance to our destination. That defeated the purpose of the bicycle all together, as another last-last-mile commute had to be made.

Artistic render of a ZaiBike along the iconic Peranakan houses in Singapore

Enter, ZaiBike. Instead of constraining our bicycles to fixed docking stations, we decided to let the bicycle do its thing – to bring you from your current location, directly to your destination. We do this by installing an IoT device onto ordinary bicycles that allows the user to rent and return the bicycle, and the operator to track every bicycle in the fleet. Parking is no longer reliant on physical infrastructure, freeing the user to park directly at his destination.

ZaiBike’s first smart lock prototype

The user would use his/her mobile device to scan a QR code on the bicycle, which authenticates with our servers, then unlocks or returns the bicycle.

Of course, many foreseeable problems arose from this idea. The biggest problems revolved around the ability to track the bicycle’s position. The most common way of tracking any object’s location is using Global Positioning System (GPS). However, inaccuracies in a GPS can go up to 50m in built up areas, and are unable to reliably track the height of the bicycle, necessary when determining parking spots in urban areas. This made tracking using an on-board GPS or the user’s mobile GPS unreliable, causing both operators and users difficulty in finding a bicycle.

Vandals can easily exploit this idea by spoofing their mobile’s GPS (when an on-board GPS is not used), or park the bicycle in their apartment in high rise buildings to hog it.

Evidently, this problem is seen in the systems running this model.

We had to make a decision: did we want to run the risk of losing our bicycles, or build the tech to solve the problems?

As engineers, we decided to go for the latter and developed various geo-fencing and machine learning techniques that could be used to determine parking locations. Our master plan was to use the commuting data from our bicycles to determine the optimal location where users should park the bicycles in order to maximise user convenience (shortest distance to a bicycle from their location/destination), and minimize operator’s cost (least number of parking spots to enable cheaper redistribution and infrastructure). Our idea became a system that could evolve from a free-floating model to a station-based model.

Alas, in 2017, we found that such a route made little business sense in the eyes of the industry. Other free-floating systems decided to go for the former option: to make the bike as cheap as possible, and risk the location-sensing problem. Since these companies were not responsible for the negative externalities from losing their bicycles (clean up of discarded and lost bicycles), they decided it was more profitable to replenish lost bicycles than to secure them. Some would say it was an unfortunate case of over-engineering the problem, and perhaps I agree in hindsight. Perhaps good engineering couldn’t solve everything, after all.

I would like to thank my teammates May Ying, Pornthip, Yong Siang, Chang Tat, Li Zhen, Kelly, Xin Lin, Xiao Xue, and Yu Hui in this project.

Interested to know more? Have questions? You can contact me here.

Leave a Reply

Your email address will not be published. Required fields are marked *