Name: 人工智能如何提高新加坡港口的效率 (How AI made Singapore’s ports more efficient)
Uploaded: 2025-10-07T18:56:40.000Z
Duration: 9 min 24 s

Every minute is precious in global trade.

全球貿易中，每一分鐘都非常寶貴。

基礎被動語態

Which is why here at the Port of Singapore, one of the world's busiest shipping hubs, no time is wasted as operations run 24,7 all year round.

這也是為什麼在新加坡港，這個全球最繁忙的航運樞紐之一，營運全年無休，分秒不浪費。

Still, not everything runs like clockwork amidst activity.

不過，忙碌之餘，並非所有事情都能順利進行。

A silent crisis has been gripping the haulier industry in Singapore, which is a relatively small country.

在相對較小的國家新加坡，一個無聲的危機正悄悄吞噬著貨運承攬業。

We have a lot of trucks that are operating within Singapore, almost like close to 3,000 or more.

我們有將近 3,000 輛或更多的卡車在新加坡境內營運。

Generally, the trucking industry globally has some of the key issues are high operational costs and high carbon emissions.

一般來說，全球卡車運輸業面臨的主要問題是高營運成本和高碳排放。

These are attributed to the sub optimal use of resources as well.

這也歸因於資源的使用不夠理想。

The biggest one being manpower shortages in the aging workforce, be it the drivers who are driving the trucks, controllers as we call the planners who are planning the trips for for the drivers.

其中最大的問題是人力短缺，以及高齡化勞動力，無論是開卡車的司機，或是我們稱之為規劃師的調度員，他們負責為司機規劃行程。

Trucking is a fragmented industry, which means it still relies on manual processes.

卡車運輸業是一個分散式產業，這意味著它仍然依賴手動流程。

But what if there was a more efficient way to move goods from port to destination?

但如果有一種更有效率的方式，能將貨物從港口運送到目的地呢？

With rapid urbanization and rising consumer demands, supply chains are being forced to evolve.

隨著都市化快速發展和消費者需求不斷增加，供應鏈正被迫演變。

This alliance is hoping to help drive that evolution with AI powered halyard planning and location services intelligence.

這個聯盟希望透過人工智慧驅動的貨運承攬規劃和位置服務智慧來推動這項演變。

PSA has been focusing a lot on terminal operations for multiple decades.

PSA 在碼頭營運方面已經專注了數十年。

So in the last about four to five years we've also started expanding our digital solutions into the community space which is beyond the terminal where we saw that just optimizing what happens inside the terminal is not sufficient to look at the logistics industry or supply chain as a whole.

所以在過去的四到五年裡，我們也開始將我們的數位解決方案擴展到社群空間，也就是碼頭以外的區域，因為我們發現光是優化碼頭內部的營運，不足以全面看待物流業或供應鏈。

Optitruck, to put it very simply, is a transport management solution which addresses the trip planning, the trip execution and making sure that we account for all the jobs that we undertook and all the way up to invoicing and billing.

Optitruck，簡單來說，是一個運輸管理解決方案，它處理行程規劃、行程執行，並確保我們能處理所有承接的貨運，一直到發票開立和計費。

HERE technologies is a location technology and platform company.

HERE technologies 是一家位置技術和平臺公司。

What we're using is something called tour planning.

我們正在使用一種叫做行程規劃的技術。

There are many variables to define that tour.

有很多變數需要定義這個行程。

What is this AI model or the machine learning model does to find the best route in that planning phase.

這個人工智慧模型或機器學習模型在規劃階段會找出最佳路線。

So when a container or the truck has to go from point A to point B or point B to point C, the algorithm is able to efficiently predict what is the best way to move.

是以，當貨櫃或卡車必須從 A 點移動到 B 點，或從 B 點移動到 C 點時，演算法就能有效預測最佳的移動方式。

The partnership with HERE actually started off with a meeting to share some of the pain points in our industry and the initiatives that we are working on.

和 HERE 的合作，其實是從一場會議開始，分享我們產業的痛點以及我們正在推動的計畫。

For a complex operations like this, one of the fundamental problem is the efficiency.

對於像這樣複雜的營運，一個根本的問題就是效率。

So our shared vision is to work with PSA to provide efficiency, reduce cost or the downtime where there are empty containers or empty trucks.

所以我們共同的願景是與 PSA 合作，提供效率、降低成本，或是減少貨櫃或卡車空駛時的停機時間。

那麼，這項技術是怎麼運作的呢？

只要輕輕一點，行程就啟動了。

Every delivery is made up of an intricate network of ports, hauliers and logistics systems.

每一趟遞送都由港口、貨運承攬業和物流系統組成的複雜網絡構成。

As of early 2025, the system had onboarded 400 trucks, acquiring close to 20% of Singapore's market share.

截至 2025 年初，該系統已導入 400 輛卡車，市佔率接近新加坡的 20%。

Each truck now handles about 10 to 12 deliveries daily.

每輛卡車現在每天大約處理 10 到 12 趟遞送。

Before the system's launch, up to 35% of trips were empty, wasting fuel, labour and maintenance costs.

在系統上線前，高達 35% 的行程是空駛，浪費了燃料、人力和維護成本。

Unproductive trips effectively means that the truck is not carrying a container.

無生產力的行程，就等於卡車沒有載運貨櫃。

It's like a taxi is moving without a passenger.

就像計程車沒有載客一樣在跑。

Ever since we have implemented Optitruck Sometime in late 2022, we have seen that they managed to reduce their empty trips almost by 50%.

自從我們在 2022 年底導入 Optitruck 系統以來，我們看到他們將空趟次減少了將近 50%。

Cutting empty trips isn't just about logistics.

That's where automation and resource sharing come in.

這就是自動化和資源共享的用武之地。

One is automated scheduling and asset pooling.

So automated scheduling largely focuses on addressing the productivity and efficiency of the controllers who are actually planning the trips.

自動排程主要著重於提升實際規劃行程的調度員的生產力和效率。

There are manpower shortages and the aging workforce.

有著人力短缺和高齡化勞動力。

So what happens five years down the line when there aren't many uptakes for the controllers?

所以五年後，當越來越少人願意擔任調度員時，會發生什麼事？

Can the system be able to do it automatically based on data and AI?

系統能不能根據數據和 AI 自動完成？

So there could be cases where I have trucks, but

所以可能會出現這種情況，我有貨車，

I don't have enough jobs to do, whereas my partner or my friend has a lot more jobs to do than the number of trucks they have.

但我沒有足夠的工作，而我的夥伴或朋友擁有的工作遠多於他們的貨車數量。

Why don't we get together, pool our resources, and

我們為什麼不聚在一起，整合我們的資源，

in doing so effectively, we need lesser number of trucks.

這樣做的話，我們需要的貨車就會比較少。

But optimizing resources is just one piece of the puzzle.

但優化資源只是拼圖的一小部分。

Real efficiency comes from something even more powerful.

真正的效率來自於更強大的東西。

Imagine a concept of a ride hailing app where the truckers are actually being allocated jobs and loads based on their proximity, their job type, the overall competency type.

想像一下叫車 App 的概念，貨車司機根據他們的距離、工作類型、整體能力類型來分配工作和貨載。

So tool planning makes use of the underlying layer of location data, which is geocoding attributes.

所以工具規劃利用了底層的位置資料，也就是地理編碼屬性。

Point of interest, a normal human eye on a picture can only look at 30, 40 attributes.

興趣點，一般人肉眼看一張圖，只能看 30、40 個屬性。

But our technologies and what we call within the industry is a digital reality index.

但我們的技術，也就是業界所說的數位實境指數。

It goes beyond 900 attributes on one static picture.

它超越了單一張靜態圖片的 900 個屬性。

That is the sheer power to convert those snippets on a picture to usable data.

這就是將圖片中的片段轉換成可用數據的強大能力。

So a particular sign becomes your speed limit, road sign.

所以一個特定的標誌就變成了你的速限、路標。

So all this attributes are blended in to give the most accurate result possible.

所以所有這些屬性都會被融合，以提供最精確的結果。

Now, when the truck hits the road, then a lot of dynamic things happens, right?

現在，當卡車上路時，就會發生很多動態的事情，對吧？

You have an accident, traffic congestion, you need to take a detour.

你發生事故、交通壅塞，需要繞道。

These efficiency gains rely on precise digital mapping.

這些效率的提升依賴於精確的數位地圖。

So how reliable is this approach when faced with the unpredictability of everyday roads?

那麼，當面對日常道路的不可預測性時，這種方法有多可靠？

We have something called unimap methodology, which is the live injection of fresh data coming in from billions of probes, active traffic, government agencies, a Lot of, you know, feeds that comes and gets injected into the overall engine.

我們有一種稱為「unimap methodology」的方法，也就是將來自數十億個探針、即時交通、政府機構以及許多其他來源的最新數據即時注入，這些數據會被注入到整個引擎中。

The tool is actually designed to give those ETA ATDs in the most correct manner because it directly affects your cost, your serviceability, your user experience and overall ecosystem.

這個工具的設計宗旨是能最準確地提供預計抵達時間 (ETA) 和預計出發時間 (ATD)，因為這直接影響到你的成本、服務水準、使用者體驗以及整體生態系統。

According to PSA, scaling up Optitruck could cut 10 million kg of CO2 emissions annually, the same impact as planting 300,000 trees.

根據 PSA 的說法，擴大 Optitruck 的應用，每年可減少 1000 萬公斤的碳排放，這相當於種植 30 萬棵樹。

Singapore is committed to lowering emissions with a target of 60 million tonnes of CO2 equivalent by 2030.

新加坡承諾到 2030 年將碳排放量減少 6000 萬噸二氧化碳當量。

By 2050, the vision is even more ambitious.

到 2050 年，這個願景更加雄心勃勃。

Our contribution to net zero and the sustainability is to reduce the carbon footprint because it's going to consume less fossil fuel when it is being efficient.

我們對淨零排放和永續發展的貢獻是減少碳足跡，因為當效率提高時，它將消耗更少的化石燃料。

We are seeing lots of EVs also coming into the picture.

我們也看到許多電動車 (EVs) 進入市場。

These carbon emission savings are actually not for us, but it's for our customers.

這些碳排放的節省，其實不是為了我們，而是為了我們的客戶。

On top of the sustainability initiatives that we have undertaken, like electrification of our equipment, we also help our customers by building digital platforms to help them achieve such savings as well.

除了我們已經採取的永續發展措施，像是設備電動化，我們也透過建置數位平臺來協助客戶達成相同的節省。

Singapore launched its next generation tuas Port in September 2022 and expansion will continue over the next two decades with the aim of making it the world's largest, largest fully automated port terminal.

新加坡在 2022 年 9 月啟用了下一代大士港，未來二十年將持續擴建，目標是成為全球最大、全自動化的港口碼頭。

By the 2000 and 40s, Tua's port will be able to handle a capacity of 65 million containers annually, nearly double the volume of all Singapore ports in 2021.

到了 2040 年，大士港每年將能處理 6500 萬個貨櫃的運量，幾乎是 2021 年新加坡所有港口總量的兩倍。

As we scale up Tuas port and go into full capacity operations by 2040, we're also looking into building our PSA supply chain hub which is focusing on cargo operations.

隨著我們擴大興建大士港，並在 2040 年達到全面營運，我們也正著手打造我們的 PSA 供應鏈樞紐，專注於貨物營運。

So beyond container trucks, we're also looking at other kinds of cargo movements.

所以除了貨櫃卡車，我們也在關注其他類型的貨物運輸。

For example, like smaller trucks, they have very unstructured and very fragmented market where people can have one lorry to hundreds of lorries, but

例如，小型卡車的市場非常零散且分散，有些人可能只有一輛貨車，有些人則有數百輛，

但情況依然是，非常、非常手動。

They don't use any planning or transport management system.

他們沒有使用任何規劃或運輸管理系統。

So if we fix that particular angle on visibility, how it improves overall port efficiency, that's our goal.

所以，如果我們能解決這個能見度的問題，進而提升整體港口效率，那就是我們的目標。

The AI technology is going to be everywhere.

Be it an operation which is like maritime medicine, healthcare, finance.

無論是航運、醫療保健或金融等營運。

What this is going to do is deal with more and more data points.

這將能處理越來越多的數據點。

As we are evolving further, we'll be able to handle such volume, such scale and of course there will be very different conditions might come in.

隨著我們持續演進，將能應付如此龐大的規模，當然也可能出現各種不同的情況。

But we believe with our experience dealing with such changes or the evolution of technology over four decades, those experience will come very, very handy.

但我們相信，憑藉我們四十年來處理這類變革或技術演進的經驗，這些經驗將會非常、非常有幫助。