Connect with us

Technology

Perceptron: Risky teleoperation, Rocket League simulation, and zoologist multiplication

Published

on

Research in the field of machine learning and AI, now a key technology in practically every industry and company, is far too voluminous for anyone to read it all. This column, Perceptron (previously Deep Science), aims to collect some of the most relevant recent discoveries and papers — particularly in, but not limited to, artificial intelligence — and explain why they matter.

This week in AI, researchers discovered a method that could allow adversaries to track the movements of remotely-controlled robots even when the robots’ communications are encrypted end-to-end. The coauthors, who hail from the University of Strathclyde in Glasgow, said that their study shows adopting the best cybersecurity practices isn’t enough to stop attacks on autonomous systems.

Remote control, or teleoperation, promises to enable operators to guide one or several robots from afar in a range of environments. Startups including Pollen Robotics, Beam, and Tortoise have demonstrated the usefulness of teleoperated robots in grocery stores, hospitals, and offices. Other companies develop remotely-controlled robots for tasks like bomb disposal or surveying sites with heavy radiation.

But the new research shows that teleoperation, even when supposedly “secure,” is risky in its susceptibility to surveillance. The Strathclyde coauthors describe in a paper using a neural network to infer information about what operations a remotely-controlled robot is carrying out. After collecting samples of TLS-protected traffic between the robot and controller and conducting an analysis, they found that the neural network could identify movements about 60% of the time and also reconstruct “warehousing workflows” (e.g., picking up packages) with “high accuracy.”

Teleoperations

Image Credits: Shah et al.

Alarming in a less immediate way is a new study from researchers at Google and the University of Michigan that explored peoples’ relationships with AI-powered systems in countries with weak legislation and “nationwide optimism” for AI. The work surveyed India-based, “financially stressed” users of instant loan platforms that target borrowers with credit determined by risk-modeling AI. According to the coauthors, the users experienced feelings of indebtedness for the “boon” of instant loans and an obligation to accept harsh terms, overshare sensitive data, and pay high fees.

The researchers argue that the findings illustrate the need for greater “algorithmic accountability,” particularly where it concerns AI in financial services. “We argue that accountability is shaped by platform-user power relations, and urge caution to policymakers in adopting a purely technical approach to fostering algorithmic accountability,” they wrote. “Instead, we call for situated interventions that enhance agency of users, enable meaningful transparency, reconfigure designer-user relations, and prompt a critical reflection in practitioners towards wider accountability.”

In less dour research, a team of scientists at TU Dortmund University, Rhine-Waal University, and LIACS Universiteit Leiden in the Netherlands developed an algorithm that they claim can “solve” the game Rocket League. Motivated to find a less computationally-intensive way to create game-playing AI, the team leveraged what they call a “sim-to-sim” transfer technique, which trained the AI system to perform in-game tasks like goalkeeping and striking within a stripped-down, simplified version of Rocket League. (Rocket League basically resembles indoor soccer, except with cars instead of human players in teams of three.)

Rocket League AI

Image Credits: Pleines et al.

It wasn’t perfect, but the researchers’ Rocket League-playing system, managed to save nearly all shots fired its way when goalkeeping. When on the offensive, the system successfully scored 75% of shots — a respectable record.

Simulators for human movements are also advancing at pace. Meta’s work on tracking and simulating human limbs has obvious applications in its AR and VR products, but it could also be used more broadly in robotics and embodied AI. Research that came out this week got a tip of the cap from none other than Mark Zuckerberg.

Simulated skeleton and muscle groups in Myosuite.

Simulated skeleton and muscle groups in Myosuite.

MyoSuite simulates muscles and skeletons in 3D as they interact with objects and themselves — this is important for agents to learn how to properly hold and manipulate things without crushing or dropping them, and also in a virtual world provides realistic grips and interactions. It supposedly runs thousands of times faster on certain tasks, which lets simulated learning processes happen much quicker. “We’re going to open source these models so researchers can use them to advance the field further,” Zuck says. And they did!

Lots of these simulations are agent- or object-based, but this project from MIT looks at simulating an overall system of independent agents: self-driving cars. The idea is that if you have a good amount of cars on the road, you can have them work together not just to avoid collisions, but to prevent idling and unnecessary stops at lights.

Animation of cars slowing down at a 4-way intersection with a stoplight.

If you look closely, only the front cars ever really stop.

As you can see in the animation above, a set of autonomous vehicles communicating using v2v protocols can basically prevent all but the very front cars from stopping at all by progressively slowing down behind one another, but not so much that they actually come to a halt. This sort of hypermiling behavior may seem like it doesn’t save much gas or battery, but when you scale it up to thousands or millions of cars it does make a difference — and it might be a more comfortable ride, too. Good luck getting everyone to approach the intersection perfectly spaced like that, though.

Switzerland is taking a good, long look at itself — using 3D scanning tech. The country is making a huge map using UAVs equipped with lidar and other tools, but there’s a catch: the movement of the drone (deliberate and accidental) introduces error into the point map that needs to be manually corrected. Not a problem if you’re just scanning a single building, but an entire country?

[embedded content]

Fortunately, a team out of EPFL is integrating an ML model directly into the lidar capture stack that can determine when an object has been scanned multiple times from different angles and use that info to line up the point map into a single cohesive mesh. This news article isn’t particularly illuminating, but the paper accompanying it goes into more detail. An example of the resulting map is visible in the video above.

Lastly, in unexpected but highly pleasant AI news, a team from the University of Zurich has designed an algorithm for tracking animal behavior so zoologists don’t have to scrub through weeks of footage to find the two examples of courting dances. It’s a collaboration with the Zurich Zoo, which makes sense when you consider the following: “Our method can recognize even subtle or rare behavioral changes in research animals, such as signs of stress, anxiety or discomfort,” said lab head Mehmet Fatih Yanik.

So the tool could be used both for learning and tracking behaviors in captivity, for the well-being of captive animals in zoos, and for other forms of animal studies as well. They could use fewer subject animals and get more information in a shorter time, with less work by grad students poring over video files late into the night. Sounds like a win-win-win-win situation to me.

Illustration of monkeys in a tree being analyzed by an AI.

Image Credits: Ella Marushenko / ETH Zurich

Also, love the illustration.

Technology

India’s Tata Motors wants to sell 50,000 EVs by end of fiscal year

Published

on

Mumbai-based automaker Tata Motors wants to sell 50,000 electric vehicles by the end of the fiscal year ending March 31, the company’s chairperson Natarajan Chandrasekaran said during a shareholders’ meeting on Monday.

In the 2023/24 period, Tata — which produces passenger cars, trucks, vans, coaches, buses, luxury cars, and construction equipment — aims to hit 100,000 EV sales, according to Chandrasekaran, as reported by Reuters.

The push towards EVs follows a national plan to ensure that up to 30% of total passenger car sales in India are electric by 2030, up from about 1% today. E-scooters and e-bikes will account for 80% of two-wheeler sales, up from 2% today. Given the Indian government’s high import duties on EVs, getting citizens to make the switch to electric will largely depend on the success of local production.

After attempting to bring its EVs to the Indian market, Tesla appears to have abandoned efforts to set up a factory in the country. Tesla usually has a “try before buy” approach to moving into new markets — it imports vehicles to see how sales go before investing the time and money in building a regional factory. Transport minister Nitin Gadkari said Tesla was welcome to build a factory in the country, but that it won’t allow the automaker to bring in vehicles from China to sell and service, so Tesla hasn’t moved forward with those plans.

Tata currently sells three EV models, including Nexon EV, Tigor EV and the newest Nexon EV Max. Unlike the path many U.S. automakers have followed of building new EV production lines from the ground up, Tata says it’s able to keep costs down for the Indian consumer by repurposing a successful internal combustion engine model, the Nexon, and outfitting it with a battery pack. The Nexon starts at around $19,000, which isn’t exactly cheap for the average Indian driver, but is certainly within the range of the country’s upper-middle class.

Tata commands 90% of India’s electric car sales, and appears to be on track to reach its goal of selling 50,000 EVs by March 2022. The automaker’s June sales results show 45,197 total units sold, out of which 3,507 were electric — the most Tata has ever sold, and up 433% from 658 last year.

Chandrasekaran was optimistic about the trajectory of Tata’s performance this fiscal year with the overall supply situation, including that of semiconductors, improving and stabilizing.

Continue Reading

Technology

Without a clear ask, your pitch deck is useless

Published

on

You’ve brushed off your Keynote skills, you’re giddy that you’re finally going to be able to start paying yourself a living wage, and you are excited to start pitching your startup’s next round of funding to your investors. It’s heady times, for sure, but hit the other pedal there for a moment, friend — you may be forgetting something.

After working with hundreds of founders on raising money — including the fantastically popular Pitch Deck Teardown series here on TechCrunch+ — there’s one slide that almost every founder gets woefully wrong. The slide is often referred to as The Ask. Or, as one investor friend calls it, the “what is my $10 million going to buy me”? slide.


TechCrunch+ is having an Independence Day sale! Save 50% on an annual subscription here. (More on TechCrunch+ here if you need it!)


The Ask is a sensitive topic to a lot of inexperienced entrepreneurs, which makes sense. Trying to right-size a funding round can be a little overwhelming, and there are a thousand different ways of building a startup. If you were successful in raising $8 million, you can do things one way. If you raised $12 million, you could perhaps launch more features of your product a little faster, or experiment more, or go after an additional market earlier. You know that. Your senior staff knows that. Your investors know that. But regardless, you need a Plan A.

What do those key metrics need to look like in order to raise not this round of funding, but your next one?

What do you need to do?

A lot of founders will tell you that they are trying to raise enough money to survive for the next 18 months. That’s probably true, but that will be true regardless of how much money you raise. A better approach is to think about what you need to accomplish to raise your next round of funding, and then work backward from there. This is probably a combination of metrics and milestones.

Metrics are the measurable parts of your business that grow and evolve over time. One of the best metrics you have is revenue, but there could be many others: the number of sales, average order value (AOV), monthly or annual recurring revenue (MRR or ARR, respectively), customer acquisition cost (CAC), customer lifetime value (LTV), daily and monthly active users (DAU and MAU), retention rate (usually expressed by its inverse, churn rate) and much more. What do those key metrics need to look like in order to raise not this round of funding, but your next one?

Milestones are also measurable parts of the business, but instead of tracking them over time, they tend to be binary: You’ve either hit a milestone or you haven’t. For startups, this could be key hires; finding the perfect, experienced CFO that can help take your company public is one major milestone a lot of companies at some point need to hit. Product launches (coming out of beta), launches in particular markets (launching only in California) and localization (launching your app in Spanish and French, for example) are also important milestones. Financial milestones are also common; the first time you make a single dollar from any customer is a huge shift in the business. When a customer, on average, starts to make you more money than it costs you to acquire them is another. For earlier-stage companies, completing a customer validation phase by talking to, say, 100 potential customers is a milestone.

When you’re raising money, you will be mapping out a set of milestones that you need to hit in order to validate your company. In addition, you’ll set a number of trigger points for metrics — hitting $1 million ARR, having 5,000 daily active users or finding a combination of customer acquisition channels that means you can acquire customers at a reasonable blended CAC, for example.

So let’s examine how to put together a great “ask” slide by ascertaining what it takes to determine how much you need to raise, how to create a specific set of goals and how to bring it all together in a coherent whole.

Continue Reading

Technology

Tech doesn’t get more full circle than this

Published

on

Welcome to Startups Weekly, a fresh human-first take on this week’s startup news and trends. To get this in your inbox, subscribe here.

Tech innovation is a cycle, especially in the main character-driven world of early-stage venture capital and copycat nature of startups.

The latest proof? Y Combinator this week announced Launch YC, a platform where people can sort accelerator startups by industry, batch and launch date to discover new products. The famed accelerator, which has seeded the likes of Instacart, Coinbase, OpenSea and Dropbox, invites users to vote for newly launched startups “to help them climb up the leaderboard, try out product demos and learn about the founding team,” it said in a blog post.


TechCrunch+ is having an Independence Day sale! Save 50% on an annual subscription here. (More on TechCrunch+ here if you need it!)


If it sounds familiar, it’s because — in my perspective — Y Combinator is taking a not-so-subtle swipe at Product Hunt, a nearly decade-old platform that is synonymous with new startup launches and feature announcements.

Y Combinator doesn’t necessarily agree with this characterization: The accelerator’s head of communications, Lindsay Amos, told me over email that “we encourage YC founders to launch on many platforms — from the YC Directory to Product Hunt to Hacker News to Launch YC — in order to reach customers, investors and candidates.”

The overlap isn’t isolated. As Y Combinator makes a Product Hunt, Product Hunt is making an Andreessen Horowitz. Meanwhile, a16z is making its own Y Combinator. Not to mention Product Hunt has investment capital from a16z and formerly went through the Y Combinator accelerator.

The strategy is more than a tongue twister, it’s a signal on what institutions think is important to offer these days (and why they’re starting to borrow more than sugar, or deal flow, from their neighbors).

For my full take, read my TechCrunch+ column, “YC makes a Product Hunt, Product Hunt makes an a16z, a16z makes a YC.”

In the rest of this newsletter, we’ll talk about Coalition, Backstage Capital and Africa’s temperature-fluctuating summer. As always, you can support me by forwarding this newsletter to a friend or following me on Twitter or subscribing to my blog.

Deal of the week

Coalition! Built by a quartet of women operators in venture, Coalition is a fund meets network that is trying to get more diverse decision-makers onto cap tables. The two-pronged approach of fund and network helps Coalition cover multiple fronts: Founders can turn to the firm for capital or the network for advice at no further dilution. Aspiring investors and advisers can turn to the firm to begin building out their portfolio, and LPs can put money into an operation that is committed to broadening diversity on cap tables, known to have economic benefits.

Here’s why it’s important: Coalition co-founder Ashley Mayer, the former VP of communications for Glossier, explained a little about the building philosophy behind the new company.

Mayer explained that she and her three co-founders saw the value of taking a “portfolio approach” to careers, basically going deep on their respective operator roles while also angel investing and eventually scout investing. Three of them previously worked in venture but left it because they missed the experience of operating. Now, they’re trying to scale a way for people to keep their day jobs and build beyond it. Coalition co-founder and Cityblock Health founder Toyin Ajayi said that “as one of few women of color leading a venture-backed company, I feel a deep obligation to hold the door open for others.”

Coalition investors (left to right): Cityblock Health co-founder Toyin Ajayi, Tribe AI co-founder Jackie Nelson, Umbrella co-founder Lindsay Ullman, Glossier VP of Communications Ashley Mayer

Image Credits: Coalition

When do layoffs matter? Trick question — always

This week on Equity, we spoke about Backstage Capital laying off a majority of its staff, weeks after pausing any investments in new startups. The workforce reduction, which impacted nine of Backstage Capital’s 12-person staff, was due to a lack of capital from limited partners, per fund founder Arlan Hamilton.

Here’s why it’s important: Backstage Capital has invested in over 200 startups built by historically overlooked entrepreneurs, while Hamllton herself has invested in more than two dozen venture capital funds. Despite having impact, no single firm can be immune from the difficulties of venture (or growing in an environment full of macroeconomic and cultural hurdles). Below is an excerpt of my story.

Without more support, it becomes difficult to close shop on new investments, bring more assets under management and bring more follow-on investments, Hamilton said.

“Somebody asked me, ‘why don’t you have more under management?’” she said during the podcast. “You gotta ask these LPs, you gotta ask these family offices, you gotta ask these people who ask me, ‘how can I be helpful,’ and I say ‘invest in our fund,’ and I never hear from them again.”

one chess pawn on a green elevated platform, with one on lower pink platform. startups and Market downturns

Image Credits: Jordan Lye (opens in a new window) / Getty Images

Africa charts its own course

TC’s Dominic-Madori Davis and Tage Kene-Okafor wrote a story about how the downturn is playing out in Africa, essentially answering why we should all be tuning into the continent’s activity this summer.

Here’s why it matters: Africa’s venture capital totals weren’t too shabby in the first quarter, but investors think that it may just be a reporting delay. If most of the deals were finalized before high interest rates, the war and inflation, experts say, we may see an economic downturn soon start affecting developing markets. The story doesn’t stop there; I’d read more to see what Tiger Global tells us and how August is shaping up to be a key month of movement. 

Arrows on the African landscape pointing up and down

The summer could decide this year’s fate of the African funding landscape.

Across the week

Seen on TechCrunch

OK, whose rocket just hit the moon?

This co-worker does not exist: FBI warns of deepfakes interviewing for tech jobs

Formerly rich NFT buyers party through the pain

Robinhood almost imploded during the GameStop meme stock chaos

FTX says no active talks to buy Robinhood

Seen on TechCrunch+

Your startup pitch deck needs an operating plan

3 questions for the startup market as we enter Q3

Disclose your Scope 3 emissions, you cowards

What’s a fintech even worth these days?

Until next time,

N

Continue Reading

Trending