n TECHNOLOGY 20 September 2024 www.drivesncontrols.com AN AMERICAN ROBOTICS start-up has pioneered a new kind of AI-powered motion planning system that needs 1,000 times less computational time than existing programming tools, and allows robots to be deployed in less than a day, rather than a month. Jacobi Robotics, founded by a group of roboticists and AI researchers from UC Berkeley and Carnegie Mellon University, has attracted $5m of seed funding. Until recently, robot motion planning – controlling robots’ joints to manoeuvre their end-of-arm tools, while avoiding obstacles – was almost never used in commercial settings because of the complexity of the algorithms. Jacobi says it has pioneered a new kind of AI-powered motion planning technology that results in the fastest-ever generation of real-time, collision-free trajectories, with calculations taking less than 1ms. The company argues that recent advances in AI, combined with labour shortages, mean that industrial robots have reached an inflection point. Shorter production runs and new applications enabled by advances in AI require robots to adapt to changes. They are now being reprogrammed every six months, rather than every six years. Jacobi says that early adopters of its technology have achieved 95% reductions in deployment times and 24% savings in project costs. The company reports that there has been “significant demand” for its motionplanning technology, with agreements already secured with manufacturers in the US, robotics suppliers including Formic, and Fortune 500 companies worldwide in sectors including electronics and consumer packaged goods. “Just a few years ago, machine learning was limited to a small group of experts and researchers,” says Jacobi’s CEO and cofounder, Max Cao. “Today, any software developer can train a production-ready machine-learning model by following an online tutorial. Similarly, in a few years, software engineers will be implementing production-ready robot applications using Jacobi. “The strong demand we’ve seen from the market underscores how programming for industrial robots is ripe for a massive shift,” he adds. “The new financing will enable us to quickly ramp Jacobi’s capacity to meet initial and new demand fulfilment.” Jacobi has also announced a palletiser – the first in a series of AI-powered applications based on its technology. It points out that more than $400bn worth of US trade is exported annually on pallets, driving a rapidly growing demand for robotic palletisers that can stack cases onto pallets for storage and shipment. www.jacobirobotics.com US start-up claims that it can program robots 20 times faster BECKHOFF AUTOMATION HAS released a generic library that provides robot functions for PLC users. Called TcRobot (TwinCat Robot), it is an open-source library that simplifies the integration of any robotic arm with Beckhoff control systems by providing a common set of functions for PLC developers, thus cutting the time needed to integrate new arms. The library uses a uniform set of commands that simplify development using standardised interactions. It will be familiar to users of Beckhoff’s PLC motion library, and leverages the power of the company’s wider ecosystem. Any TcRobot interface with existing Beckhoff technologies can be re-used for all robotic assets. The code can also be re-used, avoiding the need for repetitive coding and simplifying updates. New robots are said to be easy to add and integrate. Beckhoff says that different robots will become interchangeable and easier to integrate and control using the same generic TcRobot APIs. A common interface reduces the learning curve for different robot APIs and avoids the need to rewrite existing code to work with other robots. Previous “islands of automation” can become part of larger ecosystems. Beckhoff also argues that the new library will also simplify learning curves for manufacturers, and cut development times for PLC programmers. Future capabilities planned for TcRobot include support for the IIoT, condition monitoring, palletising, recipe building, and coordination with Beckhoff’s TwinCat Vision technology, including analytics and event logging. TcRobot is available on an MIT opensource license, allowing it to be used and modified without consent or attribution. Users can add their own IP to the library. www.beckhoff.com/en-gb Open-source library makes it easier to integrate robot arms Jacobi Robotics says that its motion-planning technology can achieve 95% reductions in programming times compared to traditional methods. Beckhoff’s TcRobot library supports a variety of robots and communications protocols
RkJQdWJsaXNoZXIy MjQ0NzM=