Spencer Deng uses robots and AI to transform logistics

A conversation with AmCham HK

Spencer Deng uses robots and AI to transform logistics

Founded in 2014, Atlanta-based Dorabot – named for the time-traveling, robotic earless cat Doraemon in the Japanese manga series of the same name – in short order has gained clients for its automated warehouse solutions from United Parcel Service (UPS), Deutsche Post AG (DHL.DE) to A.P. Moller – Maersk A/S (Maersk-B.Co), IKEA and BASFSE (BAS.DE) and many others. It partners with high-end Japanese robotics companies Fanuc Corporation (FANUY) and Yaskawa Electric Corporation (YASKY) to custom-build robotic arms to meet customer demands, while its own DoraSorter uses complex machine learning, computer vision and motion trajectory planning developed by the company. The DoraSorter robot uses a high-speed barcode scanner, high-definition camera and a custom rack system to deliver items weighing up to 6.8 kg to 100 or more destinations. AmChamHK e-Magazine catches up with co-founder Spencer Deng in Saudi Arabia, where he is exploring new opportunities for the firm. Currently it has locations in Atlanta, Shenzhen, Brisbane and Hong Kong.

How I got started

I thought the whole logistics industry or package industry had huge potential to use robotics. UPS was the best logistics company in the world, with both mature systems and innovation. At that time I was already in the US, UPS offered me a role bridging their APAC office and headquarters. I didn’t say no.

I was based in Atlanta 90% of the time, then different parts of the US. I went to Singapore and Dubai but didn’t travel to China a lot. I was at UPS for three years. The reason I wanted to go to UPS was because I wanted to understand what kind of tech would transform the whole industry. When I saw the operations, it gave me ideas, but I wouldn’t call it robotics or Artificial Intelligence (AI). UPS sent me to China and after a year I started Dorabot.

I started by using tech to solve the pain points in UPS warehouses and distribution centers. You could call it robotics with the arms doing the loading and unloading. UPS was our customer from the very beginning and then became our investor. I left but had plenty of opportunities to go back.

I founded Dorabot with Hao Zhang, our tech guy and Chief Technology Officer, who spent many years in robotics and has a PhD in artificial intelligence from the Hong Kong University of Science and Technology. Dandan Zhou, our other co-founder, spent more than 10 years in software development and was chief software engineer of video game maker Gameloft. He met Hao Zhang at Chaihuo Makerspace, Shenzhen’s first maker/hackerspace. He leveraged that to build software systems for robotics. We all had a good friend in common, and we talked for hours and days to get started with the partnership. The time was pretty short. When you have an idea and a tech background, money was chasing you. We got our funding in one or two days.

At the beginning

We started in Shenzhen. I think it’s logical and pretty natural to build robots or a hardware related company in Shenzhen. The whole supply chain is there, and you can get all the resources you need in one day. Shenzhen also attracted talent to the city. It embraces people of different backgrounds. Nobody feels rejected. We didn’t have very complicated thinking. It was a small team, with just a few people. And Shenzhen was very welcoming. You can find your own way to innovate. It’s a very sophisticated matrix when you need support.

It was almost three years before we set up in Hong Kong. We had two kinds of opportunity. We were restructuring Dorabot into a global company, which means a BVI holding company that we could take everywhere, not just China. The first opportunity came when a pharmaceutical company client wanted to upgrade warehouses in Asia, starting with Hong Kong. We leveraged the structure for a listing. We already had a Hong Kong holding company, so that it was not just a holding company. After two years we started hiring for the Hong Kong office. Most of our business was overseas. We were still facing China, but we wanted a better location for the domestic as well as the international market. Hong Kong was the best location for our real headquarters, from 2020 to right now.

Hong Kong as our “real” headquarters

Our team is less than 300 at present, most in Shenzhen, some in Guangzhou, some in Atlanta, Tokyo and Singapore. We are building a team in Dubai and Abu Dhabi. Our research is in Hong Kong. Most of the automation companies are near Shanghai, so we go there a lot. All the integrators are there. If talking about holistic integration, eastern China is important.

How we apply AI to robotics

We differentiate our AI by different parts. One is related to robotics, using neural networks to train robots which package is A, B or C. The way they pick it up is also based on AI. We have to train the robot arms to pick and place. Sortation and palletization are terms that were used when humans were doing the work. We also have AI at the back end, dealing with data. We have a pure AI team to control logistics flow, to plan the sequence of loading items into standard containers, and the dynamic sequence, which robots to use at which time. The latter is very much on the software side.

After Covid-19

During Covid I spent most of the time within mainland China, until the end of lockdown. The market expanded until Covid reached another level. After Covid, we refocused on international markets. I’ve been flying a lot, especially in Asia-Pacific, Japan, Southeast Asia, the Middle East, even Central Asia and Africa. I think the US market is still very good, but we have to be prepared when we expand in the US again. The engineering team will be at another level from a cost perspective. The whole sector is still very encouraging. In China, we have many scenarios that we deploy. For example, we have AI loading containers for a furniture company that has different types of furniture. We can do it much faster and save 5-10% in terms of space. Different manufacturers all have storage centers that are automated. They put our robots into palletization and production lines. Our clients are both state-owned and private. We have clients in pharma, chemicals, consumer goods, beverages – it’s pretty diverse, actually.

The future of robotics

With recent tech breakthroughs like ChatGPT there are new ways to develop robots to a different level in different industries. Tesla is building humanoids. These are exciting topics that will take some years to develop but are promising. We are a global company and want to embrace change with new ways to develop robots – whether it’s hardware, software, cost savings, or more flexible units in terms of function.

The Greater Bay Area

For me it’s pretty natural. We aim at the international market and are very much leveraged on the supply chain in China. We look for the middle point. Hong Kong is in the best position. It used to be locked down. There were a lot of blockers between Hong Kong and mainland China.

Now it’s 14 minutes by high-speed rail. Customs is quite fast. It’s easy to come back and forth in the Greater Bay Area with more sophisticated networks, hardware and software that we can use. That kind of perspective is a great opportunity for us to leverage from the US and from Hong Kong in the international markets. Now we work mostly with our customers from Tuen Mun, not far from the airport. We are trying to base an office on Hong Kong Island to lock it down. The problem is my energy.

After a brief career in banking, Spencer Deng co-founded the logistics robotics startup Dorabot and became its CEO. He grew up in Shunde, Guangdong. The economics graduate from Sun Yat-sen University was an associate at JPMorgan Chase & Co. before joining United Parcel Service (UPS), where he became manager for the Asia-Pacific region. He joined Dorabot in 2015, and was featured in Forbes China’s 30 under 30 in 2017 and 2018.