Finishing first can be a losing strategy. Rodney Brooks, co-founder of Rethink Robotics, the Boston company that pioneered the development of human-centered, collaborative robotics, may be ruminating on that right now. Earlier this month, the company closed its doors. The whole story hasn’t emerged yet but many believe the reason for closure was a combination of competition from rival Universal Robots and too few models that third-party developers had little interest in working with.
Ten years ago, when Brooks came up with his iconic, expressive robots, Baxter and Sawyer—the first robots that were programmable just by moving their arms—getting robots to do what they were supposed to was an achievement in itself. At the time, roboticists and those working on the fledgling internet of things or IoT had no idea how these systems could be used securely and at scale; let alone how they could share the data they collect.
Fast forward 10 years and the advocates of decentralized technologies would have robot and IoT developers believe that they will provide the necessary authentication, validation and security they've been dreaming about.
At present IoT developers have produced lots of objects with a single point of failure. Think of all the 'smart' objects in your home: virtual assistants, smart lights, smart locks, security cams. If the WiFi goes down, these smart objects become little more than sleek paperweights. In theory, a decentralized network can adjust to failures like Wifi to help eliminate or minimize security or functional risks. But are these decentralized systems ready? And, more importantly, are IoT developers and other non-blockchain technologists now ready to listen?
“Roboticists think of blockchain as a linear data structure, and they think that’s imminently unscalable,” says Lawrence Lundy, a partner at Outlier Ventures, a VC firm which invests in companies at the convergence of blockchains with artificial intelligence, robotics and the IoT. “From the outside, all you see is that it takes 10 minutes to complete a transaction and, of course, that doesn’t solve your problem. But within the industry, we see the solutions—sharding, Plasma, privacy preserving tools. There’s a whole universe of invention,” he says. “So we end up in a world where we can see how fast this is developing, but others can’t.”
This state of affairs, says Lundy, is what led to blockchain-AI company Fetch pivoting from drones to developing the world’s first adaptive smart ledger to enable “autonomous economic agents” that can pay tax, take out insurance and enter into legal agreements autonomously. “The simple fact was that there was no existing infrastructure available for them to build their use case,” says Lundy. “They had to go further down the stack and develop their own protocols.”
Earlier this year, Lundy authored “The Convergence Ecosystem,” arguing the need for a framework for a Web3 paradigm, to get people to start thinking about blockchain/AI/IoT use-cases or problems that require an integrated approach.
Synergies between blockchain and AI resonate more than those involving the physical world, where things are more complex, says Lundy. For example, people are already aware that data is stored centrally, and projects such as Wibson and the AI-enabled Ocean Protocol are working to remedy this by developing decentralized data marketplaces. While the IoT industry understands the need for shared infrastructure, data sources and networks, effective decentralized solutions haven’t yet appeared.
Forecasts from Ericsson would have us believe that by 2022 some 29 billion IoT devices will be in operation, with a potential $11 trillion at stake. But before we get to this trillion dollar data dreamland, researchers keep finding security holes in IoT devices. Earlier this year, reports emerged that consumer robots, such as those being developed by Softbank, were easy targets for ransomware attackers, who could force the hapless bots to demand money, display porn in public places and cuss out customers. Solutions are needed sooner rather than later.
One glimmer of hope on the horizon is SingularityNET, which is developing a platform for future robots to be able to communicate and process transactions securely and at scale.
But, these ideas are, as Lundy puts it, very much at the theoretical stage. However, there are others, such as swarm robotics, where development has moved apace. Swarm robotics is the use of numerous, autonomous robots, which coordinate to accomplish a task. In this field, immediate use-cases—in construction, logistics and moving goods around—have become apparent. But swarm robotics also presents a massive coordination challenge, DLT, again, is emerging as a front-runner for a solution.
Swarming towards a solution
The world of swarm robotic systems is very polarized, with some researchers working exclusively on theoretical projects and others investigating areas such as Amazon deliveries and self-driving cars, says Eduardo Castello Ferrer, a postdoctoral fellow at MIT Media Lab. “There’s no one trying to bridge these communities; there’s no one trying to bridge the gap between academia and real-world industry,” he says. “So there are fundamental problems in swarm robotics systems that aren’t being tackled.”
Since 2016, Ferrer has been combining swarm robotics and blockchain technologies to develop security, behavior and business models for distributed systems. He believes that blockchain is a viable solution to some of the problems within swarm robotics operations and can make the technology more secure, autonomous, flexible and even profitable.
By way of example, he points to faulty robots (or “malicious members”) as a problem that can affect a swarm and even cause it to combust. In order to recognise these malicious actors—which could be conveying erroneous information—Ferrer and his collaborators created “a type of lie detector” to highlight inconsistencies: “It’s a type of reputation mechanism,” he explains. This mechanism would show up in a ledger as a “history book” to alert the operator and the rest of the swarm to instances where it might not be wise to trust the information that an individual robot is providing to the group.
“We don’t want to limit the authority of these devices,” says Ferrer. “As we connect more and more robots and make them communicate with each other, the complexities aren’t going to go away.”
Although confident that his are the first projects in this nascent field, Ferrer says that others are now springing up.
Russian startup, AIRA, for instance, is developing smart contracts for autonomous drones, which, it says, would dispense with the need for air traffic control and could be used for monitoring forest fires, detecting and preventing illegal activity and inspecting solar panels, among other things.
Last summer, AIRA took part in a project to showcase a “smart, sustainable city,” featuring an autonomous P2P mesh network, on Russia’s Volga river. According to the organizers’ blog, the drones were paired with the Ethereum-based Robonomics platform. On receipt of orders about the area to be monitored, the drones secured smart contracts and planned their route. At mission’s end, they automatically transmitted a hash file to the interplanetary file system (IPFS), a protocol for storing hypermedia, containing data on their mission.
According to Ferrer, other researchers are working on various aspects of combining AI, IoT and robotics with DLT, on developing secure standards and solutions to deal with the data that will be generated by millions of connected objects. He foresees that, in the future, it will be common for blockchain-enabled robot swarms to be hired out to organizations such as city councils, who can use them for purposes such as maintenance of street lights and repairing potholes on motorways. And, in a bid to expand the blockchain-robotics space, he and his team are putting on the first “Symposium on Blockchain for Robotics Systems,” at MIT in Boston, on December 5, 2018.
And Lawrence Lundy is optimistic too. “If we look at innovation as a metric, then I’m super-bullish,” he says. “Most of the technological challenges will be dealt with in a 12-18 month period—especially the ones for IoT, robotics and M2M [machine-to-machine communication].”
The reason Lundy is so optimistic has much to do with the speed at which new innovation can permeate the blockchain industry, with most new code being open source. But having the technology available to put the decisions of robots and AI’s on a ledger is just the beginning because then we’ll need to face up to bigger challenges. The siloed nature of research in emerging technologies is one issue but the biggest will be that current regulatory headaches—such as the distinction between utility and security tokens--will seem like a walk in the park when we come to legislate on the decision making properties of autonomous objects, and how much power they should be allowed to wield.