It's easy to lose sight of the fact that robotics is as much a software problem as a hardware problem. Naturally, programming is overshadowed by the allure of mechatronics, but without the right software solution you're little more than an expensive paperweight. The path to widespread adoption of robotics is fraught with unforeseen challenges that could ultimately hinder real-world use. There are many problems that require software solutions.
Jacobi Robotics was founded in 2022 with a specific problem in mind: singularity. Confusingly, this word has a very different meaning in robotics than it does in the world of Ray Kurzweil's predictions of AI progress.
In the world of robots, the concept is much more subtle, but requires actual knowledge of the category to fully understand. It's the kind of term you rarely encounter outside of research papers. But this is a very real problem with real-world implications.
“Singularity is the Achilles heel for industrial robots,” Jacobi points out. “For repetitive tasks where a robot blindly performs the same actions over and over again, a robot can be programmed to avoid singularities through the tedious task of manually fine-tuning the robot's path over many weeks. However, in many robotic applications, small changes in material or thermal expansion require the robot's path to change periodically.
If you're at all familiar with robotics hardware, you've probably heard the term “degrees of freedom” to refer to a robotic arm with, say, 6 or 7 degrees of freedom. This refers to the system's joints and the axes along which those joints can move. A singularity is a point in space where the robot cannot move. When that happens, humans usually have to intervene to get things running again.
The name Jacobi Robotics comes from the Jacobian matrix, which is named after the pioneering 19th century German mathematician Karl Jacobi. In the world of robotics, this concept refers to the relationship between joint and end effector velocities. To further simplify what I've already oversimplified, the concept and the company named after it have to do with path planning for robots.
Jacobi Robotics was founded by four UC Berkeley robotics students and professor Ken Goldberg. Goldberg is the company's chief scientist and co-founder of package sorting robot company Ambi Robotics, so he's been to the rodeo before.
First, the team is almost exclusively focused on problems with singularities, which can cause robots to stop running at unpredictable times. In the world of robotic arms, this is a big problem for key applications such as bin picking, package sorting, and palletizing. These are more or less important things that come up when talking about industrial robots.
Jacobi has been conducting pilots with selected partners. That list includes automation company Formic, as well as a large consumer electronics company that the company isn't yet ready to name (you know how this kind of thing works in the corporate world). right). According to Formic, Jacobi's approach to attacking singularities has significantly reduced deployment times, even in the early stages. Indeed, for startups like Formic, it's in the best interest of addressing as many potential issues as possible during the deployment process, rather than sending out technicians after the fact.
In addition to Goldberg, the company's founders include CEO Max Cao, CPO Yahav Avigal, Chief Architect Lars Berscheid, Chief Robotics researcher Jeff Ichnovsky (also an assistant professor at CMU's Robotics Institute) is on the list. Jacobi said he raised his $1 million pre-seed in early 2023 and is currently focused on raising the right seed to bring its solution to market. Current investors include Swift Ventures and UC Berkeley accelerator Berkeley SkyDeck, which also joined the startup as part of a recent demo day.
The software currently provides support for many leading robotic arm vendors, including ABB, Fanu, Universal, and Yaskawa.