In San Francisco today, booking a car ride does not necessarily mean having a driver pick you up at your location. Instead, you might open the Waymo One mobile application, enter your destination, and within minutes, a silent and driverless car will pull up, and there is no one behind the wheel.
What was once a scene from the Netflix sci-fi series Black Mirror has now become a way of life there.
Waymo is a subsidiary of Alphabet, the parent company of Google, and it operates one of the world’s largest commercial robotaxi services.
The company’s robotaxis now clock more than 250,000 rides each week across cities in the United States such as Austin, Los Angeles, Phoenix and San Francisco. In April, Waymo began hitting the streets of Tokyo in Japan, one of its first international locations.
In Guangzhou, China, a similar experience unfolds with self-driving giant WeRide, where a few effortless taps on your mobile phone will summon a car that arrives quietly, without a driver. All you need is a passcode and you are on your way.
WeRide’s autonomous technology also powers larger vehicles such as its fully autonomous eight-seater robobus.
Ms Tan Shi Hui, 33, who took the robobus during a business trip to the southern Chinese city last year, found the ride surprisingly ordinary.
“I felt quite calm throughout. The acceleration and deceleration were smooth, there were no sudden jerks, and even the turns felt gentle,” the independent consultant said, adding that she did not even feel “a hint of anxiety” when on the bus.
The large window panels on all sides allowed her to see clearly what was happening outside on the road, which made the journey feel reassuring, Ms Tan recalled.
Similarly, Mr Wu Linhan and his wife, Ms Amandine Honvault, both 28, tried Baidu Apollo’s self-driving taxis when they were in Guangzhou back in 2023.
Baidu Apollo is the autonomous driving platform developed by Baidu, a leading Chinese technology company.
“It felt like stepping into the future, especially when you see the car turning without anyone’s help,” Mr Wu, a freelance video and photo editor, said.
Ms Honvault, a software engineer, said that they both felt safe because the taxi was not going fast or weaving through heavy traffic.
“We knew someone was checking our itinerary and double-checking that we don’t have issues,” she added.
Hearing of such developments, where integrated autonomous vehicles (AVs) have become a part of everyday transport, it would seem that Singapore is lagging behind, even though it is often viewed as a global leader in smart technology.
Since 2015, Singapore has been facilitating on-road AV trials, but these have been primarily at controlled sites such as closed campuses or shuttle loops at Ngee Ann Polytechnic or Resorts World Sentosa.
Under current regulations, AVs can only be operated as part of authorised trials. Private ownership or public use of driverless cars remains prohibited.
Even so, experts told CNA TODAY that Singapore’s more deliberate and cautious approach towards AV testing, though more “demanding”, prioritises safety and builds up confidence for users with the technology.
Mr Daniel Chow, principal at consulting firm Arthur D Little Southeast Asia, said: “Singapore is not the first to deploy autonomous vehicles, but it may be one of the first to get it right.”
Mr Chow, who advises his firm’s clients on business growth among other areas, added that Singapore’s “slowness” in deploying AVs stems from various factors, including the need to build confidence in operating them on bus routes, ensuring public transport operators can train workers effectively and assuring commuters that the technology has been thoroughly tested before going mainstream.
Indeed, in response to queries from CNA TODAY, the Land Transport Authority (LTA) said that while countries such as the US and China have deployed fleets of AVs as robotaxis, safety remains the key focus in Singapore.
“Safety is paramount and our priority is to ensure the local operating environment is suitable to facilitate their deployment in a safe manner, while also addressing issues related to public acceptance of such technology,” the authority added.
It also said that its first phase of AV deployment will focus on fixed-route shuttles such as road-sweeping and logistics AVs.
“LTA will ensure all safety and public acceptability issues are adequately addressed first, before moving on to other more complex use cases like robotaxis,” it added.
Mr Soh Ming, founder and managing director of Volt Auto, the authorised distributor of Dongfeng automobiles in Singapore, said: “The cautious approach makes sense, because in a dense city like Singapore, you don’t just roll out autonomous technology without being absolutely sure it’s safe.
“There’s no room for error.”
Dongfeng, one of the largest car manufacturers in China, has been actively developing autonomous driving technology with trials involving robotaxis, autonomous trucks and smart mobility solutions as part of its push into next-generation transport.
“We are moving slowly, but we are moving with intent. When Singapore does scale AVs, it will be with a system that is tested, integrated and trusted,” Mr Soh added.
HOW FAR AV HAS COME AROUND THE WORLD
Self-driving technology has made significant progress in recent years around the world, reaching a baseline maturity that enables it to operate on many urban roads without human drivers.
These autonomous vehicles can follow traffic rules, change lanes, react to traffic lights and avoid obstacles or pedestrians. However, AVs still need human oversight in complex scenarios such as extreme weather, unpredictable human behaviour or unusual road conditions.
Currently, most AVs that exist are classified at Level 3 or Level 4 autonomy, based on the six levels of driving automation established by the Society of Automotive Engineers (SAE), recognised as the global standard.
Level 0 refers to full human control of the vehicle. Level 5, the highest, represents full autonomy, where the vehicle can operate on any road and under all conditions without any human intervention.
With AV technology advancing rapidly, some countries such as China, Japan and the US have embraced autonomous vehicles extensively and are scaling up their commercial operations.
Besides Waymo in San Francisco, Amazon subsidiary Zoox is preparing to expand production of its purpose-built autonomous vehicles, which feature bidirectional driving and no steering wheel.
Similarly, in China, car manufacturer Xpeng has launched the P7+, an electric vehicle featuring the company’s XNGP autonomous driving system.
Unlike rival systems, Xpeng’s does not rely on LiDAR (Light Detection and Ranging), a technology that uses laser pulses to create precise 3D maps of the environment, but which can be costly and bulky.
Instead, Xpeng’s system relies on a combination of cameras, ultrasonic sensors and radar to navigate complex urban conditions.
WeRide has expanded rapidly beyond China, launching robotaxi services in Abu Dhabi in partnership with ride-hailing firm Uber. It plans to extend operations to Dubai and 15 other cities globally over the next five years.
The company has also received approval in China to operate fully unmanned paid ride-hailing services in Beijing, and it holds permits for passenger rides, highway driving and remote driverless operations in Guangzhou.
Competitor Pony.ai is not far behind; at the Auto Shanghai 2025 exhibition recently, the company unveiled its seventh-generation autonomous driving system, which allows for 100 per cent mass production of robotaxis to automotive standards.
In China, Pony.ai has reduced production costs of its AV systems by 70 per cent, bringing the company closer to profitability and making mass robotaxi deployment more commercially viable.
SINGAPORE’S DRIVE TO THE FUTURE
In Singapore, various testbeds have been rolled out across the island over the years. These include autonomous shuttle trials on Sentosa Island, campus-based deployments at the National University of Singapore and One-north, and small-scale commercial pilots such as NTUC FairPrice Group’s collaboration with logistics partner Zelos.
To address the decreasing number of truck drivers, FairPrice Group Supply Chain Business began deploying the Zelos Z10, the first fully driverless autonomous vehicle designed for the logistics sector, to transfer pallets of goods between its warehouses in Joo Koon in October 2024.
Since April 2024, WeRide Robobus has been deployed at selected hours as a pilot service to ferry passengers around Resorts World Sentosa.
Over at Ngee Ann Polytechnic, MooVita, a software provider of autonomous driving technology, has been providing autonomous shuttle services within the campus since January 2023.
To boost productivity and improve the working conditions of its employees, cleaning service Chye Thiam Maintenance has deployed the WeRide S1 Robosweeper to sweep the outdoor areas of The Esplanade and the WeRide S6 Robosweeper for road sweeping in the Marina Coastal Drive area since October 2024.
Singapore’s approach to AV integration is supported by the Centre of Excellence for Testing and Research of Autonomous Vehicles-NTU (CETRAN), established in August 2016.
CETRAN serves as a dedicated facility for testing and certifying AVs, ensuring that safety assessments are conducted before deployment on public roads.
Last November, during the Singapore International Transport Congress & Exhibition 2024, Minister for Transport Chee Hong Tat talked about Singapore’s progress in the AV space, reiterating that safety is the nation’s top priority when assessing the deployment of the technology for commercial use cases.
Even though Singapore would like to “move more quickly”, the current “calibrated and data-driven approach” is best given its dense urban environment and to ensure that the different elements of the ecosystem are ready, he said.
As for robotaxis, Mr Chee said the authorities remained open to exploring them, but he added that the ones offering point-to-point services are more complex to deploy safely compared with AVs that operate with fixed routes.
For now, trials will focus on AVs operating on defined routes, such as logistics vehicles and minibuses. However, if fixed-route robotaxis prove commercially viable, LTA would be open to exploring those as well, Mr Chee added.
“As technology continues to mature and public confidence grows, we will be able to scale up deployment across more routes and more use cases, including the possibility of full-fledged robotaxis in future,” he said.
In January this year, LTA issued a Request for Proposal (RFP) to pilot autonomous public bus services. The initiative aims to assess the technical feasibility of AV technology for public bus services and gain insights into operational aspects at both service and fleet levels.
The pilot will start with smaller buses of at least 16 seats on bus service 400, which runs around the Shenton Way bus terminal, and bus service 191, which loops from Buona Vista Bus terminal to Media Circle in One-north.
These areas were chosen for their shorter and simpler routes.
The RFP states that the buses will initially operate with bus drivers onboard as safety operators, with the possibility of moving to remote supervision after demonstrating reliability for at least six months.
LTA plans to procure six autonomous buses for a start, with operations set to begin in mid-2026 for an initial period of three years. Depending on their performance, up to 14 more buses may be acquired to expand the pilot to two more services.
Other AV initiatives involving Singapore companies include ComfortDelGro’s first robotaxi pilot programme in Guangzhou, which was launched in partnership with Pony.ai.
The two-year pilot, initiated in March this year, aims to develop and refine capabilities for autonomous vehicle technology operations and fleet management, with the goal of large-scale deployment in China and other international markets.
ComfortDelGro, Singapore’s largest taxi operator, said in response to queries from CNA TODAY that the robotaxis have been “well-received” by the public in Guangzhou, with consistent daily numbers.
So far, the pilot has experienced zero at-fault traffic incidents and zero passenger incidents, it added.
“The technology has demonstrated that it can adapt and manoeuvre well in mixed traffic and weather conditions with minimal virtual assistance, ensuring a smooth and safe ride for passengers,” ComfortDelGro said.
“The lessons learnt would be useful for application in Singapore and other parts of the world when regulations permit in the future.”
On whether ComfortDelGro will introduce its robotaxis to Singapore as well, the company said that for the robotaxis to be commercially viable on a larger scale, the technology will have to be mature enough for safe mixed-environment driving, supported by strong regulation and public trust.
“We have seen successful case studies of robotaxis on the road in China, San Francisco and other parts of the world. However, for it to take off on a more significant scale globally, it would require wider public acceptance and a stronger push at the policy level.”
“A CAUTIOUS APPROACH” BETTER FOR SINGAPORE
While other cities might be racing ahead to integrate autonomous vehicles into their public transport systems, Singapore’s “slow” approach reflects the city’s dense and highly complex environment, experts said.
Mr Alex Tang, head of Xpeng’s international division of sales and service in Singapore, said: “Bringing AV technology to a densely populated and highly regulated environment like Singapore requires a deliberate, step-by-step approach.”
It is crucial to consider the unique characteristics of Singapore’s urban landscape, including its high traffic density, diverse mix of road users who include those on personal mobility devices and the weather conditions here, he added.
Dr Dan Chia, senior lecturer at the Singapore Institute of Technology (SIT), said Singapore’s many pilot programmes and “step-wise” approach are also due to its lack of “greenfield spaces” as compared with other cities such as Guangzhou and San Francisco.
Greenfield spaces are undeveloped lands ideal for unrestricted design and testing.
Dr Chia, who researches AVs and intelligent transport systems, added: “China and the US have greenfield spaces to play in, where areas and surroundings can be clearly marked without too much public interaction. But in Singapore, it’s difficult, since we lack a true greenfield to test AV tech.
“A step-wise approach may seem a little bit slower compared to overseas cities, but this reduces the danger and unwanted accidents, given the more complex scenarios in Singapore.”
Although the island’s density might seem like a roadblock to adopting autonomous vehicles, Mr Chow from Arthur D Little Southeast Asia said that it is “our proving ground” and works to the country’s advantage in its foray into AV technology.
“Autonomous vehicles in Singapore have to navigate tight, highly dense and complex environments … if you can make autonomous technology work here, you can make it work in most urban environments.”
Agreeing, Mr Tang from Xpeng said the city’s well-regulated roads and high compliance with traffic rules offer a “conducive environment” for deploying and refining autonomous systems.
Dr Chia said that deploying public buses is a more challenging endeavour than robotaxis, due to real-life scenarios and “occlusions” that are difficult to test for, as well as the high expectations people have for public transport.
Occlusions are situations where a vehicle’s sensors are blocked or cannot clearly detect objects, people or other vehicles.
“When you talk about public roads, the speed of operation is different and the environment is more complex because you will have more occlusions,” Dr Chia said, citing situations such as when cyclists appear unexpectedly.
He noted that Singapore’s cautious approach is also because AV buses face more challenging demands than robotaxis, which carry just a few passengers. Buses must handle larger passenger volumes and, if standing passengers are allowed, smooth acceleration and braking become critical to avoid injuries.
Furthermore, even if the technology and testing proceed as planned, large-scale deployment of autonomous vehicles requires an “ecosystem” to ensure long-term business viability.
ECOSYSTEM WITH MANY STAKEHOLDERS
Developing an AV transport system does not rely on technology alone – it requires “many moving parts that need to move together in concert”, Mr Chow said.
Agreeing, Mr Tang said Singapore presents unique challenges that would require significant and sustained investments, “not just in physical infrastructure, but also in digital systems and urban design that support safe and efficient AV operations”.
“It’s a confluence of different factors. The government can certainly try to orchestrate (the developments), like it currently does with electric vehicles and electrification,” Mr Chow added, noting that the upcoming RFP from LTA is a “step in the right direction”.
“Going forward, you’ll need an ecosystem approach where all of the various actors and players within the transport and mobility ecosystem invest capital, invest time and invest resources to make it happen.”
Even with successful trials, Dr Chia cautioned that deploying AV buses is “not so straightforward”. It can be costly and complex, requiring specialised expertise, extensive training and high overheads, making it a business that needs “deep pockets”.
“This business model is very different from typical bus services provided by operators like Tower Transit or under the SBS model,” he said.
“Existing bus drivers do not have the technical expertise to operate AV buses, and the technological requirements are far beyond the current baseline.”
Dr Chia noted that AV buses are far more expensive to purchase than conventional models and may not turn a profit for years, even if they run 24 hours a day, assuming that maintenance costs are manageable.
Professor Zhao Jinhua, the co-lead principal investigator of Mens, Manus and Machina, an interdisciplinary research group at the Singapore-MIT Alliance for Research and Technology, said that scaling autonomous vehicles safely remains “a formidable challenge”.
“As AV fleets grow, the ability to maintain them efficiently becomes crucial.”
Prof Zhao said this extends beyond software and includes the physical upkeep of vehicles and robotic systems, which are needed to perform tasks such as cleaning, charging, diagnostics and repairs of the fleet for minimal downtime.
“Automating these operational tasks is a foundational step toward making AV deployment not just technologically feasible, but economically sustainable.”
Source : https://www.channelnewsasia.com/today/big-read/big-read-self-driving-vehicles-5116971