Robust.AI, Three years ago, AI was established for—well, the exact purpose was unclear. Or rather, a few groups of individuals—the Robust—were certain of what. The AI team was quiet and kept to themselves. However, it was the founding team that caught our attention because it comprised some highly qualified individuals from SRI and Alphabet as well as some well-known personalities in academia: Gary Marcus, Henrik Christensen, and Rodney Brooks.
A recent press release provided clarification on what Robust is. We had a brief conversation with Brooks to straighten out all the details after AI had been working on it but also raised some concerns.
The key points from the press release are as follows:
Robust.AI, the company revolutionizing how robots work for people, has introduced Carter, a hardware product prototype, as well as Grace, its first software suite product. To enhance interactions and collaboration between humans and machines, Grace and Carter integrate robotics, AI, and human-centered design. These technologies will be used in the warehousing sector, where more than 80% of warehouses run manually and manpower shortages continue to affect the global supply chain.
Using the Grace software suite, people and robots may dynamically coordinate in any warehouse with any workflow. Work can be divided up between humans and robots using the software, which also allows for distribution and adaptation as needed. It provides situational awareness, including semantic and human perception as well as mapping and localization using cameras, to make robots capable in environments where people move about and work. Through a no-code interface, warehouse managers will be able to quickly alter the behaviour, integrations, and workflows of an entire fleet of robots.
Robust developed when Grace was in development. The Autonomous Mobile Robots (AMR) market is currently slow to develop, deploy, and accepts for a variety of applications, according to AI. Robust was created as a result of this insight. Carter, a Collaborative Mobile Robot, is the first piece of hardware created by AI (CMR). Carter offers adaptable automation for handling materials in warehouses and other settings. Carter increases engagement and productivity of people by working around them without altering the surroundings, enabling smooth coordination.
Grace and Carter accomplish a number of things that other AMR firms currently do, like working in the same environments as busy, squishy humans and a mobile robot that can move objects, so the natural question is how precisely Grace and Carter differ from other AMRs. It’s vital to notice that the mobile robot is free from any manipulators.
It was crucial to let people know that we weren’t working on manipulators despite persistent rumours to the contrary, state Brooks. That’s the drawback of operating in secret for so long, but Rodney Brooks expresses genuine enthusiasm for what Robust.AI is doing and, more importantly, for how different it is from other systems, stating that “both Grace and Carter we believe provides features that we believe are not really available on other systems.”
Grace, let’s start with the stack of software. According to Brooks, this method eliminates the requirement to joystick a robot around to create an initial map by allowing a user to create a VSLAM (visual simultaneous localization and mapping) model of the world by just walking through their workspace. Joysticking about is not really challenging, but it is normally something that AMR companies do as part of a deployment and Robust. AI aims to minimise that adoption hurdle by doing rid of the demand for a special setup procedure. The Grace system then assigns duties to people and robots while users identify locations and regions to direct the robots where to go. Without needing to do any reengineering, Brooks asserts that the entire system may be deployed in the cloud. “We aim to start automating the 80% of American warehouses that don’t currently have any automation.”
Carter is a portable robot that sounds like “Cart-er” and is about the size of a shopping cart. It lacks lidar but is packed with cameras (16 in the present version), each of which is running a model powered by neural processors. As a result, the only data that has to be passed around is semantic data in addition to coordinates.
The most important feature of Carter, according to Brooks, is that once you hold its handlebar, it enters power-assist mode and may be easily moved in any direction. “Most other AMRs, even those with good sense, move where they choose, and the best a human can do is dance with them to intervene to stop them. The crucial point is that human workers benefit from automation and are not kept apart from it.
“[Our robots] actually see and comprehend people. We have started to create some really basic semantic judgments about what individuals will do next based on what they are now doing. Instead than acting like they are the boss of the area, our robots are respectful of people. They should be friendly with the staff members, please.” said Roger Brooks
Brooks recently used bears, goats, and dogs to illustrate the goals Robust.AI has for its system. Think of it like this:
AGVs are similar to bears in that it is preferable to keep your distance from them as getting too close to them can be harmful and, even in the best-case scenario, they are not friendly. The majority of AMRs, though, resemble goats. Although a goat is unlikely to eat you and you can probably spend time with it without getting mauled, depending on the goat, you may find that it exhibits some behaviours that bother you. The goat will act in goat-like ways and treat you mostly as a hindrance, a source of food, or something to headbutt.
Robust.AI While this is going on, AI is creating what it refers to as “Collaborative Mobile Robots,” which Brooks imagines would act more like service dogs. Instead of merely existing safely in your environment, CMRs will pay attention to what you’re doing, comprehend what you want, and work with you. Like a partner in a dance, says Brooks.
The versatility of this technology and software, according to Brooks, is a benefit, particularly in small businesses where work is always changing. Carter fits nicely in in factories where moving goods around is already done with carts. Although it is different from warehouse fulfilment, it has many of the same elements and enables Robust.AI to target a larger variety of situations.
For the record, Robust AI can’t compete with highly organised warehouses or online retailers like Amazon and Ocado. Additionally, the corporation isn’t focusing on situations that are only minimally structured, where the surroundings and workflows are solid enough that no adjustment is necessary. Robust. AI is focusing on “micro-fulfillment,” which refers to industrial processes and spaces up to 50,000 square feet.
According to Brooks, the smallest warehouse they have seen is 100 square feet. For us, that is too small. But more 10000, 20000, and 50000 square foot warehouses are being built as fulfilment centres, and these warehouses need automation systems that are simple to use. We’re attempting to implement automation with minimal work.
Brooks believes that reducing effort might result in Robust. Even for modest facilities, the AI system makes sense. Such were locations that could only require one robot. Because the first deployment makes it so difficult to justify something that small, I don’t believe many other AMR firms would even try to sell a single robot. According to Brooks, AI itself may not be interested in doing this in the long run. “We’re searching for partners who want to construct and sell these robots with our software stack and who have a significant presence in manufacturing, selling, and repairing machines. We’ll achieve a vast magnitude much more quickly in this way. We believe that we can enter this industry rapidly and it is excellent.
While the images and video serve as an excellent illustration of the kind of content that Robust.AI produces. Carter is only a prototype, and the Grace software stack, according to what we’ve been told, is also being developed for other platform. AI is planning to deploy. Many consumers are being spoken to, says Brooks, “and we anticipate having trials this year.” And Robust by the next year. The goal of AI is to have a production line set up to the point where it can market its robots.