Archives

artificial intelligence

Google research makes for an effortless robotic dog trot

As capable as robots are, the original animals after which they tend to be designed are always much, much better. That’s partly because it’s difficult to learn how to walk like a dog directly from a dog — but this research from Google’s AI labs make it considerably easier.

The goal of this research, a collaboration with UC Berkeley, was to find a way to efficiently and automatically transfer “agile behaviors” like a light-footed trot or spin from their source (a good dog) to a quadrupedal robot. This sort of thing has been done before, but as the researchers’ blog post points out, the established training process can often “require a great deal of expert insight, and often involves a lengthy reward tuning process for each desired skill.”

That doesn’t scale well, naturally, but that manual tuning is necessary to make sure the animal’s movements are approximated well by the robot. Even a very doglike robot isn’t actually a dog, and the way a dog moves may not be exactly the way the robot should, leading the latter to fall down, lock up, or otherwise fail.

The Google AI project addresses this by adding a bit of controlled chaos to the normal order of things. Ordinarily, the dog’s motions would be captured and key points like feet and joints would be carefully tracked. These points would be approximated to the robot’s in a digital simulation where a virtual version of the robot attempts to imitate the motions of the dog with its own, learning as it goes.

So far, so good, but the real problem comes when you try to use the results of that simulation to control an actual robot. The real world isn’t a 2D plane with idealized friction rules and all that. Unfortunately, that means that uncorrected simulation-based gaits tend to walk a robot right into the ground.

To prevent this, the researchers introduced an element of randomness to the physical parameters used in the simulation, making the virtual robot weigh more, or have weaker motors, or experience greater friction with the ground. This made the machine learning model describing how to walk have to account for all kinds of small variances and the complications they create down the line — and how to counteract them.

Learning to accommodate for that randomness made the learned walking method far more robust in the real world, leading to a passable imitation of the target dog walk, and even more complicated moves like turns and spins, without any manual intervention and only little extra virtual training.

Naturally manual tweaking could still be added to the mix if desired, but as it stands this is a large improvement over what could previously be done totally automatically.

In another research project described in the same post, another set of researchers describe a robot teaching itself to walk on its own, but imbued with the intelligence to avoid walking outside its designated area and to pick itself up when it falls. With those basic skills baked in, the robot was able to amble around its training area continuously with no human intervention, learning quite respectable locomotion skills.

The paper on learning agile behaviors from animals can be read here, while the one on robots learning to walk on their own (a collaboration with Berkeley and the Georgia Institute of Technology) is here.

OctoML raises $15M to make optimizing ML models easier

OctoML, a startup founded by the team behind the Apache TVM machine learning compiler stack project, today announced that it has raised a $15 million Series A round led by Amplify, with participation from Madrone Ventures, which led its $3.9 million seed round. The core idea behind OctoML and TVM is to use machine learning to optimize machine learning models so they can more efficiently run on different types of hardware.

“There’s been quite a bit of progress in creating machine learning models,” OctoML CEO and University of Washington professor Luis Ceze told me.” But a lot of the pain has moved to once you have a model, how do you actually make good use of it in the edge and in the clouds?”

That’s where the TVM project comes in, which was launched by Ceze and his collaborators at the University of Washington’s Paul G. Allen School of Computer Science & Engineering. It’s now an Apache incubating project and because it’s seen quite a bit of usage and support from major companies like AWS, ARM, Facebook, Google, Intel, Microsoft, Nvidia, Xilinx and others, the team decided to form a commercial venture around it, which became OctoML. Today, even Amazon Alexa’s wake word detection is powered by TVM.

Ceze described TVM as a modern operating system for machine learning models. “A machine learning model is not code, it doesn’t have instructions, it has numbers that describe its statistical modeling,” he said. “There’s quite a few challenges in making it run efficiently on a given hardware platform because there’s literally billions and billions of ways in which you can map a model to specific hardware targets. Picking the right one that performs well is a significant task that typically requires human intuition.”

And that’s where OctoML and its “Octomizer” SaaS product, which it also announced, today come in. Users can upload their model to the service and it will automatically optimize, benchmark and package it for the hardware you specify and in the format you want. For more advanced users, there’s also the option to add the service’s API to their CI/CD pipelines. These optimized models run significantly faster because they can now fully leverage the hardware they run on, but what many businesses will maybe care about even more is that these more efficient models also cost them less to run in the cloud, or that they are able to use cheaper hardware with less performance to get the same results. For some use cases, TVM already results in 80x performance gains.

Currently, the OctoML team consists of about 20 engineers. With this new funding, the company plans to expand its team. Those hires will mostly be engineers, but Ceze also stressed that he wants to hire an evangelist, which makes sense, given the company’s open-source heritage. He also noted that while the Octomizer is a good start, the real goal here is to build a more fully featured MLOps platform. “OctoML’s mission is to build the world’s best platform that automates MLOps,” he said.

Using AI responsibly to fight the coronavirus pandemic

The emergence of the novel coronavirus has left the world in turmoil. COVID-19, the disease caused by the virus, has reached virtually every corner of the world, with the number of cases exceeding a million and the number of deaths more than 50,000 worldwide. It is a situation that will affect us all in one way or another.

With the imposition of lockdowns, limitations of movement, the closure of borders and other measures to contain the virus, the operating environment of law enforcement agencies and those security services tasked with protecting the public from harm has suddenly become ever more complex. They find themselves thrust into the middle of an unparalleled situation, playing a critical role in halting the spread of the virus and preserving public safety and social order in the process. In response to this growing crisis, many of these agencies and entities are turning to AI and related technologies for support in unique and innovative ways. Enhancing surveillance, monitoring and detection capabilities is high on the priority list.

For instance, early in the outbreak, Reuters reported a case in China wherein the authorities relied on facial recognition cameras to track a man from Hangzhou who had traveled in an affected area. Upon his return home, the local police were there to instruct him to self-quarantine or face repercussions. Police in China and Spain have also started to use technology to enforce quarantine, with drones being used to patrol and broadcast audio messages to the public, encouraging them to stay at home. People flying to Hong Kong airport receive monitoring bracelets that alert the authorities if they breach the quarantine by leaving their home.

In the United States, a surveillance company announced that its AI-enhanced thermal cameras can detect fevers, while in Thailand, border officers at airports are already piloting a biometric screening system using fever-detecting cameras.

Isolated cases or the new norm?

With the number of cases, deaths and countries on lockdown increasing at an alarming rate, we can assume that these will not be isolated examples of technological innovation in response to this global crisis. In the coming days, weeks and months of this outbreak, we will most likely see more and more AI use cases come to the fore.

While the application of AI can play an important role in seizing the reins in this crisis, and even safeguard officers and officials from infection, we must not forget that its use can raise very real and serious human rights concerns that can be damaging and undermine the trust placed in government by communities. Human rights, civil liberties and the fundamental principles of law may be exposed or damaged if we do not tread this path with great caution. There may be no turning back if Pandora’s box is opened.

In a public statement on March 19, the monitors for freedom of expression and freedom of the media for the United Nations, the Inter-American Commission for Human Rights and the Representative on Freedom of the Media of the Organization for Security and Co-operation in Europe issued a joint statement on promoting and protecting access to and free flow of information during the pandemic, and specifically took note of the growing use of surveillance technology to track the spread of the coronavirus. They acknowledged that there is a need for active efforts to confront the pandemic, but stressed that “it is also crucial that such tools be limited in use, both in terms of purpose and time, and that individual rights to privacy, non-discrimination, the protection of journalistic sources and other freedoms be rigorously protected.”

This is not an easy task, but a necessary one. So what can we do?

Ways to responsibly use AI to fight the coronavirus pandemic

  1. Data anonymization: While some countries are tracking individual suspected patients and their contacts, Austria, Belgium, Italy and the U.K. are collecting anonymized data to study the movement of people in a more general manner. This option still provides governments with the ability to track the movement of large groups, but minimizes the risk of infringing data privacy rights.
  2. Purpose limitation: Personal data that is collected and processed to track the spread of the coronavirus should not be reused for another purpose. National authorities should seek to ensure that the large amounts of personal and medical data are exclusively used for public health reasons. The is a concept already in force in Europe, within the context of the European Union’s General Data Protection Regulation (GDPR), but it’s time for this to become a global principle for AI.
  3. Knowledge-sharing and open access data: António Guterres, the United Nations Secretary-General, has insisted that “global action and solidarity are crucial,” and that we will not win this fight alone. This is applicable on many levels, even for the use of AI by law enforcement and security services in the fight against COVID-19. These agencies and entities must collaborate with one another and with other key stakeholders in the community, including the public and civil society organizations. AI use case and data should be shared and transparency promoted.
  4. Time limitation:  Although the end of this pandemic seems rather far away at this point in time, it will come to an end. When it does, national authorities will need to scale back their newly acquired monitoring capabilities after this pandemic. As Yuval Noah Harari observed in his recent article, “temporary measures have a nasty habit of outlasting emergencies, especially as there is always a new emergency lurking on the horizon.” We must ensure that these exceptional capabilities are indeed scaled back and do not become the new norm.

Within the United Nations system, the United Nations Interregional Crime and Justice Research Institute (UNICRI) is working to advance approaches to AI such as these. It has established a specialized Centre for AI and Robotics in The Hague and is one of the few international actors dedicated to specifically looking at AI vis-à-vis crime prevention and control, criminal justice, rule of law and security. It assists national authorities, in particular law enforcement agencies, to understand the opportunities presented by these technologies and, at the same time, to navigate the potential pitfalls associated with these technologies.

Working closely with International Criminal Police Organization (INTERPOL), UNICRI has set up a global platform for law enforcement, fostering discussion on AI, identifying practical use cases and defining principles for responsible use. Much work has been done through this forum, but it is still early days, and the path ahead is long.

While the COVID-19 pandemic has illustrated several innovative use cases, as well as the urgency for the governments to do their utmost to stop the spread of the virus, it is important to not let consideration of fundamental principles, rights and respect for the rule of law be set aside. The positive power and potential of AI is real. It can help those embroiled in fighting this battle to slow the spread of this debilitating disease. It can help save lives. But we must stay vigilant and commit to the safe, ethical and responsible use of AI.

It is essential that, even in times of great crisis, we remain conscience of the duality of AI and strive to advance AI for good.