Tag: Emerging-Technologies

Posted on

Aurora Hydrogen raises $10M, but will its process decarbonize or facilitate tar sand exploitation?

A number of startups have cropped up to tackle the challenge of making hydrogen cheap and accessible for industrial users, including the latest, Aurora Hydrogen.

The startup announced a $10 million Series A yesterday led by Energy Innovation Capital and joined by Williams Companies, Shell Ventures, Chevron Technology Ventures and the George Kaiser Family Foundation.

Aurora said its microwave-based approach can make hydrogen using 80% less electricity than the cleanest form of hydrogen production and with less carbon emissions than the cheapest.

Posted on

Ottonomy.IO raises $3.3 million to expand network of autonomous robots for deliveries

Ottonomy.IO, a startup working on solving delivery problems using autonomous robots, has raised $3.3 million in a seed funding round as it looks to expand its market and deploy robots to existing customers.

Led by Bengaluru-based Pi Ventures, the latest funding round included participation from Connetic Ventures and Branded Hospitality Ventures. Sangeet Kumar, founder and chief executive of Uttar Pradesh-based Addverb Technologies, also joined the round.

Founded in late 2020 by Ritukar Vijay along with Pradyot Korupolu, Ashish Gupta and Hardik Sharma, New York-headquartered Ottonomy.IO develops robots that feature sensors including 3D LiDAR sensors and cameras. The company, which employs about 25 people in the U.S. and India, also writes software and AI algorithms to power the sensors.

“One of the most important problems which we are trying to solve with these autonomous delivery robots is around labor shortages,” said Vijay, who serves as the chief executive of Ottonomy.IO, in an interaction with TechCrunch. He added that due to the labor shortages, there is a substantial increase in the hourly wages of laborers — to $18 to $45 per hour from $9 to $12 — in the U.S.

“So, that’s almost a 100% hike in hourly wages, making it very difficult for enterprise customers to provide the same services to the customers they were given earlier. And what happens at the end is that customers start paying more for deliveries.”

Ottonomy.IO’s autonomous robots help address the growing demand for last-mile deliveries and cater to emerging delivery requirements including indoor and curbside deliveries. The latter is specifically a type of delivery in which the item is delivered to a place such as a parking lot — not directly to the customers’ home or office address.

“That too is very labor intensive because somebody has to bring items from a store to the curbside in the parking lot,” Vijay said.

Ottonomy.IO

Ottonomy.IO co-founders Hardik Sharma, Ashish Gupta, Ritukar Vijay, Pradyot Korupolu (from left to right)

The robots built by Ottonomy.IO use high information mapping of the serviceable delivery areas to navigate and reach the consumer delivery locations. Once they reach their destination, the robots require a unique QR code that the customer received at the time of their order to unlock the hold area and retrieve the order, the startup said.

Ottonomy.IO did some initial pilot rounds with its autonomous robots before bringing its latest model, Ottobot 2.0. This is an evolution from the early pilots and includes fully customizable modular cabins, increased customer access and directional mobility, including a crab mode that allows the robot to navigate sideways.

The proprietary robot designed by Ottonomy.IO offers better accessibility to customers, the startup claimed.

“Even a person in a wheelchair can actually access packages from the robot which is very, very important,” Vijay said.

Ottonomy.IO

Ottonomy.IO’s robots are claimed to offer better accessibility support to customers

The executive also told TechCrunch that the robot has software capability of “no GPS dependence” and “works seamlessly” for both indoor and outdoor autonomous mobility.

According to Vijay, all these design elements and technological changes make Ottonomy.IO different from the competition that includes Starship, Kiwibot, Serve Robotics, and Refraction AI, to name a few.

“If you see the other players, they’re either into indoor navigation or outdoor navigation. Given the capability of doing both opens multiple use cases for the company,” Roopan Aulakh, Managing Director, Pi Ventures, told TechCrunch.

Although the robots that Ottonomy.IO offers are developed in India, the company doesn’t consider the country its potential market.

“Our intent was not to go to solve a problem where it is not there,” said Vijay. “Otherwise, India could have been a market where grouped housing societies can still leverage autonomous deliveries, but the problem is not there because labor costs are still affordable,” he said.

Ottonomy.IO has already deployed its robots at the Cincinnati International Airport and is in talks with multiple airports in the U.S. and Europe to expand its business.

Vijay said that deploying the robots at airports was also a part of Ottonomy.IO’s go-to-market strategy to help them receive public attention.

The company said it was also working with top Fortune 500 companies in retail and restaurant industries across North America to widen its market, though it declined to identify them. Vijay said the startup is also working with a couple of fleet aggregators in Europe for last-mile deliveries. Ottonomy.IO also plans to expand to the Middle East and South America within this year, the executive said.

Ottonomy.IO plans to deploy the fresh funds to bring the robots to customers already on board — across categories including curbsides and last-mile deliveries of F&B, retail, and e-commerce packages.

“Next three to four or five months are very heavy in terms of deployments for us,” Vijay said. The startup is also looking to expand its team. “From our thesis perspective, Ottonomy.IO is a deep tech startup solving a global problem,” said Aulakh. “Developing the product in India, even manufacturing it here and selling it globally, I think is a great story, and which is what really excites us about this company.”

Ottonomy.IO raised $1.6 million before the seed funding. Some of its early investors include former Apple employees, startup founders and angels from markets including Singapore, Europe, India, the Middle East, and U.S.

Posted on

Tesla, SpaceX alums stalk energy sector’s white whale with $2 million seed round

Hydrogen has been the decarbonized energy sector’s white whale.

Wildly abundant and highly versatile, it has long had the potential to decarbonize entire sectors of the economy. From cars and trucks to planes, trains and even household boilers, the universe’s lightest gas isn’t short on possible applications.

But it is short on successful applications. Cars and trucks? Until yesterday, the last time I saw a hydrogen car was when I tested a Mercedes fuel cell B-Class … nearly 15 years ago. Planes? Not anytime soon. And homes? Japan is testing the idea, but given the difficulty of retrofitting the natural gas infrastructure to accommodate the leaky molecules, it’s unlikely to happen in the near future.

One place where hydrogen does show promise is in heavy industry, where intense heat and dense power can be hard to replicate with electric sources. Cost remains a hurdle, though.

That’s where Hgen hopes its modular electrolyzers will make an impact. The startup aims to decarbonize hard-to-crack industries like steel and ammonia production by focusing on green hydrogen that’s made using renewable power. It was founded by Molly Yang, a Tesla alum who helped lead the Supercharger, residential energy and industrial energy teams, and Colin Ho, who led actuation and power systems for Starship at SpaceX.

The company exclusively told TechCrunch that it has raised a $2 million seed round led by Founders Fund, which was joined by Fontinalis Partners, Climate Capital, Yishan Wong and a handful of other angels.

Yang said Hgen is bringing to hydrogen production Tesla’s and SpaceX’s focus on optimizing the whole widget.

Posted on

Real driverless cars are now legal in China’s tech hub Shenzhen

There are plenty of autonomous driving vehicles testing on the roads of Shenzhen today: Pony.ai, Baidu, DeepRoute, AutoX, you name it. But these vehicles are not really the unmanned vehicles tech upstarts envision for the future, as they have been required to operate with a safety driver behind the wheel.

A set of provisions introduced by the Shenzhen government is bringing the industry one step closer to a driverless future. The “Silicon Valley of China” that’s home to the likes of Huawei, Tencent, and DJI is historically known for its progressive economic policies, so it’s unsurprising that the city just became the first in China to have laid out comprehensive rules governing smart and connected vehicles.

The regulation, which is set to take effect on August 1, grants permission for autonomous driving vehicles to operate without a human in the driver’s seat — though only within areas designated by the city’s authorities.

The rules also define the thorny issue of liability. When the vehicle is equipped with a driver, the driver will “be handled” by the transportation authorities in case of traffic rule violations and incidents. But if the car is completely driverless, the owner or manager of the self-driving vehicle is subject to handling by the authorities. If the accident is a result of a defect in the connected car, the owner or manager of the car can seek compensation from the manufacturer or vendor.

Driverless robotaxis have been allowed to operate in Beijing, the capital city where events often serve as the bellwether for the rest of the country, but the permission comes in the form of case-by-case “permits” rather than being officialized in regulations as is the case with Shenzhen.

Major autonomous driving players in China have all opted for a lidar-based route instead of one that relies purely on vision tech like Tesla. Meanwhile, passenger car manufacturers are in a race to equip their latest models with advanced driving assistance tech, which is also powered by lidar. Robo vans designed to ferry things inside industrial facilities are similarly equipped with lidar.

Demand from these various fields has been a boon to domestic lidar makers like Hesai, Robosense, Livox, Innovusion as well as foreign ones such as Ouster. Luminar, the Florida-based lidar company, recently snagged a strategic investment from Ecarx, a smart car platform started by China’s auto mogul Li Shufu, the founder of Geely, which is expected to help it secure more business in China and beyond.

Luminar to invest in Geely-affiliated Ecarx, eyes China market

Posted on

Real driverless cars are now legal in China’s tech hub Shenzhen

There are plenty of autonomous driving vehicles testing on the roads of Shenzhen today: Pony.ai, Baidu, DeepRoute, AutoX, you name it. But these vehicles are not really the unmanned vehicles tech upstarts envision for the future, as they have been required to operate with a safety driver behind the wheel.

A set of provisions introduced by the Shenzhen government is bringing the industry one step closer to a driverless future. The “Silicon Valley of China” that’s home to the likes of Huawei, Tencent, and DJI is historically known for its progressive economic policies, so it’s unsurprising that the city just became the first in China to have laid out comprehensive rules governing smart and connected vehicles.

The regulation, which is set to take effect on August 1, grants permission for autonomous driving vehicles to operate without a human in the driver’s seat — though only within areas designated by the city’s authorities.

The rules also define the thorny issue of liability. When the vehicle is equipped with a driver, the driver will “be handled” by the transportation authorities in case of traffic rule violations and incidents. But if the car is completely driverless, the owner or manager of the self-driving vehicle is subject to handling by the authorities. If the accident is a result of a defect in the connected car, the owner or manager of the car can seek compensation from the manufacturer or vendor.

Driverless robotaxis have been allowed to operate in Beijing, the capital city where events often serve as the bellwether for the rest of the country, but the permission comes in the form of case-by-case “permits” rather than being officialized in regulations as is the case with Shenzhen.

Major autonomous driving players in China have all opted for a lidar-based route instead of one that relies purely on vision tech like Tesla. Meanwhile, passenger car manufacturers are in a race to equip their latest models with advanced driving assistance tech, which is also powered by lidar. Robo vans designed to ferry things inside industrial facilities are similarly equipped with lidar.

Demand from these various fields has been a boon to domestic lidar makers like Hesai, Robosense, Livox, Innovusion as well as foreign ones such as Ouster. Luminar, the Florida-based lidar company, recently snagged a strategic investment from Ecarx, a smart car platform started by China’s auto mogul Li Shufu, the founder of Geely, which is expected to help it secure more business in China and beyond.

Luminar to invest in Geely-affiliated Ecarx, eyes China market

Posted on

UK’s Oxford Quantum Circuits snaps up $47M for quantum-computing-as-a-service

Quantum computing has been making quantum leaps of progress in the last several years — going from theoretical concept to multiple testing environments, to help organizations prep for a time when quantum computers, and their unparalleled processing power, become a scaled reality. Now, UK-based Oxford Quantum Circuits is announcing £38 million ($47 million) in funding to fuel the growth of its own contribution to the space — a patented 3D processor architecture it calls Coaxmon, plus quantum-computing-as-a-service that will run on it. OQC says that this Series A is the largest to date for a UK-based quantum computing startup.

“We work at pace, and our systems are being optimized. We’ll continue to scale and reduce error rates,” said Ilana Wisby, OQC’s founding CEO, in an interview. “Our vision is seamless quantum access.”

Lansdowne Partners and The University of Tokyo Edge Capital Partners (UTEC) a deep tech fund out of Japan, are co-leading the round, with British Patient Capital, Oxford Science Enterprises (OSE) and Oxford Investment Consultants (OIC) also participating. OSE and OIC previously led a £2.2 million seed round into the startup, which began life as a spinout from Oxford University and work done there by quantum physicist (and OQC founder) Dr Peter Leek.

The plan will be to use the funding to keep hiring more talent (it’s now at 60 employees), continue improving accessibility to quantum computing for developers interested in working with it, and to continue building out its computing infrastructure, which today is based on an 8-qubit machine. And as you might guess from the investor list, it will also be using some of the funds to expand into Asia Pacific, and specifically Japan, to tap would-be customers there in financial services and beyond.

“Quantum computing promises to be the next frontier of innovation, and OQC, with its state-of-the-art Coaxmon technology, aims to integrate the forefront of modern physics into our everyday lives,” said Lenny Chin, a principal at UTEC, in a statement. “UTEC is honoured to be part of OQC’s mission of making quantum technology accessible to all and will support OQC’s expansion into Asia-Pacific through collaborations with academia including the University of Tokyo, and partnerships with Japan’s leading financial and tech corporations.”

Wisby told me that OQC actually started raising this Series A before the pandemic, back in early 2020; but it opted to shelve that process and go for grants instead to build out the company in its earlier phases.

That got OQC quite far, advancing from a 1-qubit, to a 2-qubit, then a 4-qubit, and now currently an 8-qubit machine.

The startup is also already providing services to a variety of customers who work across either OQC’s private cloud or via Amazon Braket, AWS’s quantum computing platform that also provides developers access to other quantum-as-a-service providers such as Rigetti, IonQ and D-Wave. (OQC notes that its quantum computer, named Lucy, is the first European quantum provider on Braket — a key detail for companies and quantum researchers based out of Europe who need to comply with data protection laws by keeping data and the processing of it local: this gives them a local option.)

Its customers include Cambridge Quantum, which runs its IronBridge cryptographic number generator on OQC’s computer; financial services companies; molecular dynamics researchers; government organizations and large multinationals with in-house R&D teams working on systems capable to be run on quantum machines when they are eventually spun up.

“Eventually” is the operative word here: the real promise of quantum computing is vast computing power, but there has yet to be a quantum computer built that can achieve that at scale without also producing a lot of errors.

But it seems that a lot of the hope these days is not on “if” but “when” that hurdle will be overcome. “We’re well past theory,” Wisby said.

That’s led to a big wave of both large tech players such as IBM, Amazon and Alphabet to get involved, as well as a number of smaller startups, and companies like Rigetti, IonQ and D-Wave that sit between those two poles. While there are some opting to build and sell quantum devices, the economics don’t make sense for most potential use cases, so for now the bigger efforts appear to be around quantum in the cloud: offering it as an infrastructure-free, use-as-you-need-it compute service.

Although Oxford Quantum Circuits’ 8-qubit computer is not the largest in the field, Wisby said that one reason it’s picking up users, and this investment in what has been a tough fundraising climate, is because its platform is better, in that it produces less faults than others.

“We’re all working towards larger scale processes,” Wisby said. But, she added, there is something to be said for better quality and less errors. “We have low error rates, and the funding will enable us to deliver on the next steps.”

Another major fillip in the process is the fact that regions, and countries, are looking to back leaders in the field early on to help cement their respective standing in that next generation of technology, and so backing Oxford Quantum Circuits is seen to be part of that strategy. British Patient Capital is a strategic backer in that regard: it’s the investment arm of the British Business Bank, which is a government-owned bank focused on developing business and industry in the U.K.

“Since launching the UK’s first commercially-available quantum computer, we have continued to be highly impressed with both the technical developments and also the future ambitions of OQC,” said Peter Davies, partner and head of developed markets strategy at Lansdowne Partners, in a statement. “We are very excited to be investing in this innovative and forward-thinking company.”

Posted on

Amazon extends its quantum efforts with a focus on networking

Amazon today announced a new effort in bringing quantum computing to its cloud — at least in the long term. The company today launched the AWS Center for Quantum Computing, a new research effort that aims to push forward the science and engineering of networking quantum computers together, both for building more powerful, multi-processor networks for computation and for creating secure quantum communication networks.

In recent years, Amazon and its AWS cloud computing unit made a number of major investments in quantum computing. With Amazon Braket, the company offers developers access to quantum computers from the likes of IonQ, Oxford Quantum Circuits, Rigetti and D-Wave, as well as other software tools and simulators. In addition to that, the company is also already running two more research-centric efforts: the AWS Center for Quantum Computing in Pasadena, California, which focuses on basic science like building better qubits and error correction algorithms, and the Amazon Quantum Solutions Lab, which puts an emphasis on helping enterprises prepare for the future of quantum computing.

Basically, while Braket and the Quantum Solutions Lab focus on near-term practical solutions, the Center for Quantum Computing and now the Center for Quantum Networking focus on long-term research efforts.

“To unlock the full potential of quantum devices, they need to be connected together into a quantum network, similar to the way today’s devices are connected via the internet,” the company explains in today’s announcement. “Despite not receiving the same level of attention as quantum computers, quantum networks have fascinating possible applications. One of them is enabling global communications protected by quantum key distribution with privacy and security levels not achievable using conventional encryption techniques. Quantum networks will also provide powerful and secure cloud quantum servers by connecting together and amplifying the capabilities of individual quantum processors.”

As of now, most of the research efforts around quantum networking have been at the level of start-funded research labs. Most commercial quantum computing efforts have been at the processor level (and the ecosystem around that), so it’s definitely a bit of a novelty that Amazon is now focusing on that, but now may be the time to focus on this aspect of quantum computing, especially given that quantum processing units are now starting to reach a level of maturity that would’ve been considered science fiction only a decade or two ago.

Alice&Bob, a quantum computing startup, raises $30M to launch its first fault-tolerant ‘cat qubit’ computers in 2023

Posted on

Shield AI raises $165M at a $2.3B valuation to fuel development of its military autonomous flying systems

Technology built with defense in mind is getting some significant and serious traction at the moment, spurred by world events, advances in technology, and a growing appetite from end users to invest in more innovative ways to protect themselves. In the latest development, Shield AI — which makes software and hardware for drones and other autonomous aircraft used by military and other government organizations — has raised $165 million in funding, $90 million in Series E equity and $75 million in debt.

The funding is coming in at a $2.3 billion valuation, Shield AI said. The company has been on a strong pace on that front: it follows on from a $210-$300 million Series D about ten months ago that valued the company at $1.25 billion. (It never confirmed the final amount, which was also a mix of equity and debt.)

Doug Philippone at Snowpoint Ventures led the round, with Riot Ventures, Disruptive (a returning backer; it led Shield AI’s Series D) and Homebrew (it led Shield AI’s seed round). The company’s other investors include Point72, Andreessen Horowitz, Breyer Capital, and SVB Capital.

Philippone is an interesting person to lead on this latest round: in addition to being an investor, he is also Palantir’s global defense lead, a job he’s been in for the last 14 years. This is important not least because Palantir arguably was one of the key companies to change the game for how startups, spurred by the tech boom out of Silicon Valley, both engaged and started to win defense contracts and raised huge sums from VCs to fuel that growth.

Another influential startup changing the conversation around funding defense tech is Anduril, which as we reported just the other week, is raising up to $1.2 billion (potentially more) at a $7 billion valuation. That round, we have heard, is basically now closed.

Shield AI is based out of San Diego, which you could say is a little like the Silicon Valley of the defense industry. It’s the home port of the U.S. Pacific fleet, and according to stats gathered by the city’s chamber of commerce, outside of Fairfax County, Virginia (where the Pentagon is based) greater San Diego gets more defense spending than any other place in the U.S. Shield is based there among a dozens of other major and smaller defense contractors.

And if you don’t follow the defense industry, but have at least seen or heard of Top Gun or its recently-released blockbuster sequel, you’ll know that it’s a major center specifically for aerospace development. Shield AI targets a very specific customer base that is focused around the U.S. military and its allies, but even so it speaks about what it does in terms that bring is purpose and function into context for more ordinary people.

“China’s military is Netflix; the U.S. military is Blockbuster. China is Amazon; the U.S. is Barnes & Noble. China is Tesla; the U.S. is General Motors,” writes Brandon Tseng, the president of the company who co-founded it with his brother Ryan (who is the CEO). Brandon is also a former Navy SEAL so he speaks with some authority in making these sorts of analogies.

And on the company’s home page, it describes Hivemind, its AI-based autonomous software platform, as what else? “A Top Gun for every aircraft.”

As with a lot of other companies (maybe every company) in autonomous transportation, be it in the air or on the ground, Shield AI has a mix of software and hardware that is already usable, and then products that are still in development. Some will be used in purely autonomous systems, and some in tandem with humans.

In the case of Shield AI, the company says that Hivemind and its Nova drone (or small-unmanned aircraft system, sUAS, in more formal terminology) have been in use since 2018. Ryan Tseng tells us that the specifics of exactly where and how are classified, as are most of the companies other activities, but they are part of the U.S. Department of Defense Program of Record.

It’s also working on a vertical take-off and landing (VTOL) aircraft called V-BAT that will be soon equipped with Hivemind. The software is being integrated into other aircraft, too, such as the F-16 fighter jet pictured above, where it will act as a co-pilot alongside a human, with the aim for it to be used also across F-22s, F-18s, and other models. In the meantime, Tseng said in an interview that its V-BAT craft also have been operational since 2018 around the globe.

“The DoD and international militaries are acquiring V-BAT at a rapid rate so we’re ramping production as quickly as possible,” he said — one reason for this funding. V-BAT beat out 13 competitors to win a major Navy Program of Record, he added. Its selling point is its ability to withstand challenging conditions. “The unique design and controls allow it to take off & land in high winds, on crowded flight decks, aboard moving vessels with landing zones as small as 12’ x 12’.”

The bigger strategy is to build a “swarming” capability for its devices — essentially to use a number of them in concert as a way of evading jamming technologies from adversaries. This, Tseng said, is on track for coming to market by the end of 2023 (although since a lot of what they do is classified, they may not actually make anything public until it’s already being used).

Taking both Anduril’s recent landmark round and this latest round for Shield AI, we’re in a moment right now where VCs — working themselves in a challenging financial climate — have changed their tune when it comes to backing companies in the defense space, which includes not just companies like these building military technology, but also those working in cybersecurity and other kinds of technology that helps with resilience. This could include, interestingly, alternative energy tech and of course products that can be used by more than just governments but enterprises as well.

“The fundraising climate has never been more favorable for defense technology companies,” Tseng told TechCrunch. “Supporting defense was taboo in many circles. We were rejected by many early investors because defense was considered too controversial. Today, there is growing recognition that investment in defense contributes to security, stability, and peace, all of which are foundational to a flourishing society.”

As noted by others who are investing in this space right now, or building for it, there has indeed been a noticeable shift in how people view companies like Shield AI and what they are trying to develop. That is still a challenge, though, which might be one reason why a company like Shield goes through the work of putting out messaging to people who may never actually be customers to still take in what they are trying to do.

“Many people don’t realize the scope of conflict in the world – before Ukraine, 84 million people were displaced by violence and persecution, up from 39 million in 2011,” Tseng said. “There aren’t that many opportunities to contribute to technologies that meaningfully address humanity’s great challenges – or that create the general conditions for human achievement. When you work on AI pilots for defense – you are working on the most important and disruptive defense technology of the next thirty years – and are empowering our country and allies to advance security, stability, and peace.”

That is filliped also by the fact that adversaries are also hot on the heels building their own similar systems. China is aiming for military parity by 2027 in the Pacific, Tseng pointed out, meaning they aim to exceed the U.S. by 2028. And he added that there have been reports that it is already benchmarking their prototypes against Shield AI’s pilot.

Tseng also may be biased but has a very different idea of why autonomous matters more in this context. “Waymo engineers get to build minivans that plod through the suburbs at 25 mph, we get to work autonomous fighter jets that fly 1000+ mph, dodge missiles, and find threats,” he said.

All this is spelling not just an opportunity in the business sense, but a wider one, too, for those backing Shield AI.

“Investors are flocking to quality. This round is a reflection of Shield AI’s success in creating great products, building a business with strong fundamentals, and dominant technological leadership – with an AI pilot proven to be the world’s best in numerous military evaluations,” said Philippone in a statement. “We love that they are leveraging an AI and software backbone across a variety of aircraft to deliver truly game-changing value to our warfighters. The work they are doing today is just the tip of the iceberg.”

Posted on

4 questions to ask before building a computer vision model

Eric Landau Contributor
Before Eric Landau co-founded Encord, he spent nearly a decade at DRW, where he was lead quantitative researcher on a global equity delta one desk and put thousands of models into production. He holds an S.M. in Applied Physics from Harvard University, an M.S. in Electrical Engineering and a B.S. in Physics from Stanford University.

In 2015, the launch of YOLO — a high-performing computer vision model that could produce predictions for real-time object detection — started an avalanche of progress that sped up computer vision’s jump from research to market.

It’s since been an exciting time for startups as entrepreneurs continue to discover use cases for computer vision in everything from retail and agriculture to construction. With lower computing costs, greater model accuracy and rapid proliferation of raw data, an increasing number of startups are turning to computer vision to find solutions to problems.

However, before founders begin building AI systems, they should think carefully about their risk appetite, data management practices and strategies for future-proofing their AI stack.

TechCrunch+ is having a Memorial Day sale. You can save 50% on annual subscriptions for a limited time.

Below are four factors that founders should consider when deciding to build computer vision models.

Is deep learning the right tool for solving my problem?

It may sound crazy, but the first question founders should ask themselves is if they even need to use a deep learning approach to solve their problem.

During my time in finance, I often saw that we’d hire a new employee right out of university who would want to use the latest deep learning model to solve a problem. After spending time working on the model, they’d come to the conclusion that using a variant of linear regression worked better.

To avoid falling into the so-called prototype-production gap, founders must think carefully about the performance characteristics required for model deployment.

The moral of the story?

Deep learning might sound like a futuristic solution, but in reality, these systems are sensitive to many small factors. Often, you can already use an existing and simpler solution — such as a “classical” algorithm — that produces an equally good or better outcome for lower cost.

Consider the problem, and the solution, from all angles before building a deep learning model.

Deep learning in general, and computer vision in particular, hold a great deal of promise for creating new approaches to solving old problems. However, building these systems comes with an investment risk: You’ll need machine learning engineers, a lot of data and validation mechanisms to put these models into production and build a functioning AI system.

It’s best to evaluate whether a simpler solution could solve your problem before beginning such a large-scale effort.

Perform a thorough risk assessment

Before building any AI system, founders must consider their risk appetite, which means evaluating the risks that occur at both the application layer and the research and development stage.

Posted on

Nomagic picks up $22M for its e-commerce warehouse picking robots

Robotics are playing a growing role in the world of e-commerce logistics and fulfillment — where they are seen not just as a way to speed up operations but to drastically reduce the costs of running them — and today a startup developing software and hardware specifically in the area of robot picking is announcing some funding.

Nomagic, a Polish startup that has built a robotic arm that can identify and pick out an item from an unordered selection (say, from objects in a box) and then move or pack it into another place, has raised $22 million, funding that it will be using towards both growing and expanding its business.

Nomagic’s robotic arms were first deployed to work picking up and moving small consumer electronics and related items — phones, cables, small toys — before extending to items like bagged apparel. Kacper Nowicki, the CEO who co-founded the company with Marek Cygan (CTO) and Tristan d’Orgeval (CSO), said that the plan is to add in more categories like groceries over time, reflecting changing consumer habits and what people are buying online these days. “That is the long-term goal,” he said.

The company already has a number of customers in sectors ranging from fashion, e-commerce and third-party logistics providers — one of the more prominent is Brack.ch, a Swiss-based “everything” store similar to Amazon in terms of its physical product range. And while it currently bases its tech around computer vision to identify objects and read codes, over time it is likely also to incorporate other kinds of tech such as radio-wave scanning to identify items. 

Khosla Ventures and Berlin’s Almaz Capital co-led the round with the European Investment Bank, with past backers Hoxton Ventures, Capnamic Ventures, DN Capital and Manta Ray also participating.

Nomagic last raised funding — a seed round of $8.6 million — in February 2020; and in the interim, it’s been a wild ride in the world of e-commerce.

Covid-19 led to a huge surge in online shopping, but also a reassessment of how people could work in warehouses under pandemic concerns and restrictions, and in some cases some serious reassessments of how operations were run, and a curtailing of investments to adjust to changing (and sometimes hard-hit) business conditions. Nomagic’s technology plays into all of those developments in a variety of ways.

The most obvious of these is around digital transformation, where companies are adopting robotic hardware as part of a wider update of their systems and bringing on more automation. Nomagic cites data from Research and Markets and Mordor that estimates that the the global warehouse automation market will be worth $31 billion by 2025, and that the market for piece-picking robots specifically is growing at a rate of 62.5% and will be worth $2.9 billion by 2026.

Alongside that, there is an obvious opportunity for robots to work in environments where humans might not, either because the environment is unsuitable for them and to modern labor laws (eg, no lighting, small spaces, no heating or cooling, and long hours); or because companies cannot afford that labor. Although Nomagic is also building out some hardware components, today the company focuses the bulk of its R&D on software development, which Nowicki says means that ultimately the tech will be able to work across a wide range of hardware.

Given that much larger e-commerce giants like Amazon and Ocado are investing in their own robotics technology, building third-party services that can be adopted by smaller players will be essential to letting them continue to compete. Nowicki argues that this is not about putting humans out of work but letting them focus on less repetitive tasks robots cannot handle — a factor potentially even more important for smaller organizations, with smaller staff bases and resources, to consider. This is the opportunity that investors see, too.

“An increasing number of mundane tasks will be increasingly automated by robots over the coming years,” said Kanu Gulati, partner at Khosla Ventures in a statement. “We come in early to support companies building promising technologies that are bold and impactful like Nomagic and are excited by the momentum they have demonstrated with customers.”