Back to all blogs

Singapore at Local Startup Explains Deep Tech & its Relationship with... Tomatoes?


Thu, 11/26/2020 - 12:00


23-year-old NTU Engineering student Toh Wen Qi was tasked with what seemed like an impossible task during her apprenticeship at a Deep Tech startup earlier this year: To determine the sweetness of a tomato without destroying it.

For weeks, the team she was attached to handled tomatoes of various sizes, squeezed them, juiced them, and even put them through a garlic press.

There has been some research done on this involving hyperspectral infrared imaging. So the team built a multispectral imaging system which uses near-infrared LED lights. But no matter what they did, they could not get the LED bulbs to light up. Frustrated, the team kept ramping up the power supply until it finally glowed bright red for a split second before it exploded.

“That’s how we figured that we shouldn’t expect to see near-infrared wavelengths with our eyes,” says Toh.

The team hasn’t yet come up with a solution but Toh’s experience at the startup, Polybee, is indicative of most Deep Tech startups since Deep Tech deals mostly with unexplored scientific and engineering territory; setbacks are almost a given. But for those who succeed, the reward is a stab at making history.

It is a role with high stakes and high rewards.

But First, What is Deep Tech?

A Deep Tech company is usually one that is founded on scientific discovery or meaningful engineering innovation. Because of this, its product usually has intellectual property at its core.

Doesn’t this apply to most tech companies? Not really.

Most technology companies today are based on business model innovations or simply moving physical businesses online using technology that is already available.

Deep Tech can be applied to solve world problems like famine, disease, global warming, or even traffic congestion in a city.

One example of a local Deep Tech startup is Biofourmis. This MedTech startup created a health analytics platform, Biovitals ®, which is powered by Artificial Intelligence (AI). Aside from monitoring a patient’s biovitals, it can also predict when they are going to get worse. Biovitals ® is currently being used to monitor Covid-19 patients in Australia, Hong Kong, and the UK.

Polybee, the Deep Tech startup which Toh apprenticed at, is on its way to producing autonomous drones that can perform precision pollination in indoor farming. Their goal is to provide pollination as a service in the future.

SG Government Pouring Millions into Deep Tech

Because of rapid advances in technology today, we are seeing an increasing number of Deep Tech startups.

There is obviously a lot of potential in Deep Tech and the Singapore government is willing to invest in it.

In the Budget 2020, Deputy Prime Minister Heng Swee Keat announced that the government will set aside an additional S$300 million under the Startup SG Equity scheme, to provide equity investments to Deep Tech startups with strong Intellectual Property and global market potential. S$200 million has been specifically allocated to boost support for startups in the areas of MedTech, CleanTech, and Advanced Manufacturing.

The Journey Towards Drone Pollination

Back to Polybee — producing autonomous drones to pollinate flowers won’t seem like a pretty odd goal once you know that pollination around the world is still pretty much done by hand, says Toh.

“So that's why all of these [farming] activities are being carried out in countries where the labour is cheap enough to make that profitable.”

In countries where labour is not so cheap, pollination still remains a “bottleneck” says Toh. Some flowers need to be cross-pollinated. Others must be buzz-pollinated (in nature, this is done by bees).

Pollination, as you can imagine, is extremely difficult to automate. However, that’s Polybee’s eventual goal, with the help of drones.

Since autonomous pollination requires a complex integration of different technologies, Polybee has to take small incremental steps — sort of like “technical prerequisites” says Toh, which will eventually get the team to a solution.

One of the “technical prerequisites” that Polybee is working on right now is for the technology to perform phenotyping. This is the process where you select traits from crops and breed them so that over time, you end up with a crop that has many good qualities — for instance, a sweeter taste or a longer shelf life.

Just like pollination, the process of phenotyping — eyeballing and sieving out the crops with good qualities — is largely a manual process. Polybee is attempting to automate this by creating an algorithm that can pick out crops with certain qualities just by scanning it.

“Phenotyping is a pragmatic stepping stone towards pollination in terms of product development,” says Toh.

What Working at a Deep Tech Startup is Like

It can be tedious, but Toh has also found the process to be very rewarding simply because she is challenged by the work. In fact, that was the reason why she joined Polybee through Summation, the apprenticeship programme by SGInnovate.

“I just find it challenging and fun to work on difficult problems,” she says. Polybee is also the first startup that she interned at.

The small size of the company meant that everyone had to take up more responsibilities that were outside of their original work scope. While some might balk at the messiness of it all, Toh relishes the opportunity to be constantly learning.

One of her job responsibilities was to fly drones over racks of strawberry plants but when the Covid-19 pandemic hit, Toh could not go down to the Polybee office. So she started working on projects that could be done remotely.

Like the computer vision pipeline for the phenotyping, which entailed Wen Qi creating algorithms that can detect individual chilli peppers from a photo of a pile of chilli peppers and in the process, attribute characteristics like size, length and colour. That’s pretty impressive, if you think about it.

It is work like this, which has real world impact, that Toh enjoys.

In fact, there are so many projects — with real world impact — that Polybee can do and the team — Toh included — constantly found themselves challenging the fundamental questions about their work.

Take Polybee’s work with tomatoes.

The team knows, from market research, that being able to determine a tomato’s sweetness without destroying it is very important for tomato-growers. Is this even possible? If so, is this the best problem for Polybee to take on? Why work on tomato technology? Why not work on corn phenotyping technology?

“These were all open questions that we were always discussing [especially] when our project is on the verge of tanking and we will go back to assessing the need for it.” Toh treasured this type of open communication with her Polybee colleagues. Reflecting on her apprenticeship, Toh said: “I felt free to ask about the startup’s funding, outlook, raise my doubts and concerns and provide honest feedback. In turn, I was rewarded with a lot of knowledge that I don’t think interns in any other company are typically privy to.”

Her mentor was especially cognisant of the need for the team to deliberate company issues together, even the smallest things like company brochures.

She also found herself quite at home among people who were equally driven and curious. “I think that's true for a lot of people who work in Deep Tech... They probably got into it because they were curious about the technology and want to bring it from lab to market.”

People who want to join a Deep Tech company should indulge their curiosity, she says, adding that people are naturally curious creatures, and they should harness this trait. “I think you can achieve a lot in a startup.”

At SGInnovate, we are always seeking out the best talent. Our talent programmes aim to match top talent with high-potential Deep Tech startups in the fields such as AI, Cybersecurity, Quantum Computing and more.

SGInnovate talent programmes aim to match top talent with high-potential Deep Tech startups with projects in the fields of emerging technologies such as Artificial Intelligence, Cybersecurity, IoT, Robotics, Quantum Computing and more.


Why You Need Machine Learning Operations to Achieve AI Governance

Nov 11, 2022


Helping AI Embrace Empathy

Nov 11, 2022