Motion Planning Startup Jacobi Robotics Raises $5M

Industrial robot programming startup Jacobi Robotics, which uses AI to speed up the programming of robots, has raised $5 million in seed funding.

They’ve developed a new kind of AI-powered motion planning technology that needs 1,000 times less computational time than existing programming tools. This advancement enables rapid programming of complex and large-scale robot applications, which could typically take months to complete using traditional tools.

The round was led by Moxxie Ventures with participation from Foothill Ventures, Humba Ventures, and The House Fund. Existing investors, including Swift Ventures, Berkeley SkyDeck Fund, and LDV Partners, also joined the round.

Alongside the funding announcement, Jacobi Robotics introduced its first product, the Jacobi Palletizer. This product uses Jacobi’s AI-powered motion planning technology to significantly reduce the time it takes to program robot arms for palletization tasks.

Jacobi Robotics was founded by four roboticists from the Berkeley AI Research Lab, along with Professor Ken Goldberg. Goldberg, who also co-founded the package sorting robotics company Ambi Robotics, serves as Jacobi’s chief scientist, bringing valuable experience to the team.

The startup has already completed successfully pilots with robotics solutions providers like Formic. According to Jacobi Robotics, its partners have experienced a 95% reduction in deployment time and a 24% savings in overall project costs using Jacobi’s solutions.

PS: Jacobi co-founder and CEO Max Cao will be speaking at RoboBusiness on Oct 16-17 in Santa Clara, California. He will discuss early promises of Gen AI, the main hurdles to the tech, and how Jacobi is overcoming them. If you’re attending, be sure to say hi to Max from me!

Jacobi will also have a booth at Pack Expo 2024 on Nov 3-6 in Chicago. They will be at booth N-5476 and will be giving a sneak peek of newer developments. 

Last week, I spoke with Max to learn more about Jacobi Robotics.

With so many problems in the robotics world needing better motion planning software, I asked Max why they chose palletizing as their first market.

Max shared a four-part framework that guided their decision:

1. Familiarity and Scale: They sought an application that was already widely deployed and familiar to factories, integrators, and machine builders. Palletizing, a well-established process, fit this criterion perfectly.

2. Cross-Industry Applications: Palletizing is used across multiple industries, including manufacturing, food and beverage, consumer packaged goods (CPG), and warehouses. This broad applicability made it an attractive initial target.

3. Foundation for Other Applications: Palletizing serves as a strong foundation for other material-handling tasks like depalletizing, bin picking, and sorting. This would allow Jacobi to easily expand their technology into related areas.

4. Industry Shift: The palletizing industry is transitioning from using large, four-axis industrial robots designed for high-volume, low-mix environments to collaborative robots (Cobots). Cobots are more versatile and can move between different tasks, but they require frequent reprogramming. This shift presented a significant opportunity for Jacobi’s technology.

Beyond this framework, Max and his team conducted hundreds of conversations with potential customers, gaining a deep understanding of industry challenges and validating their solutions through pilot projects.

With confidence in their ability to solve a major pain point and new funding secured, Jacobi Robotics is now well-positioned to scale their technology.

This conversation has been edited for length and clarity.

Can you share a bit about your background and how Jacobi Robotics got started?

Max Cao: Sure! We started Jacobi Robotics from the Berkeley AI research lab, where we were researching manipulation and motion planning for robot arms across various industries. Initially, we were exploring cutting-edge technologies like reinforcement learning and imitation learning for robotics. However, we quickly realized that while the technology was advancing, the gap between research and industry application was growing wider. Despite significant advancements, the adoption of automation and robotics remained low. That realization led us to found Jacobi Robotics at the end of 2022, with a mission to increase the adoption of automation by leveraging motion planning technology.

Why did you choose to focus on palletizing as your first market?

Max Cao: That's a great question. We chose palletizing because it met several criteria that made it an ideal starting point for us:

1. Familiarity and Scale: Palletizing is a well-established process, familiar to many factories and integrators. It's been around for decades, so there's a level of comfort and understanding in the industry.

2. Cross-Industry Applicability: Palletizing is used in various sectors, from manufacturing to food and beverage, and even in warehouses. This cross-industry relevance made it a broad and attractive market.

3. Foundation for Other Applications: Palletizing serves as a base for other material-handling tasks like depalletizing and sortation. Starting here allows us to expand into related areas more easily.

4. Industry Shift: The industry is shifting from traditional, industrial four-axis palletizers to more flexible collaborative robots, which require more complex programming. This shift provided an opportunity for us to showcase our technology's ability to simplify and optimize these processes.

You mentioned the increasing complexity in programming robots, especially with the shift to Cobots. Can you elaborate on that?

Max Cao: Absolutely. Traditionally, palletizing was done using large, industrial four-axis robots. These robots were programmed once for a specific task and then deployed in stable, predictable environments where the task parameters didn’t change much.

However, the industry is now moving towards Cobots, which are more flexible and can move between different production lines. This flexibility is great, but it also means that these robots need to be reprogrammed frequently to adapt to different tasks and environments. The traditional methods of programming are time-consuming and require specialized expertise, which introduces significant challenges.

Our technology simplifies this process by automating the programming through advanced motion planning. This reduces downtime and makes it easier for manufacturers to adjust to new production requirements without needing extensive manual reprogramming.

How does Jacobi Robotics’ motion planning technology work, and how does AI come into play?

Max Cao: At its core, our technology focuses on motion planning, which involves finding optimal, collision-free trajectories for robot arms in any environment. Our approach combines recent advances in AI with decades of research in traditional optimization algorithms.

We’ve developed very accurate models of robot arm dynamics and the robot workspace. These models, combined with powerful optimization techniques, allow us to calculate fast and safe robot motions.

AI comes into play by embedding these models into a deep reinforcement learning framework. This framework enables the system to generate and validate trajectories in nanoseconds, allowing us to find the optimal robot trajectories in real-time—specifically, in about one millisecond, which, to the best of our knowledge, is the fastest ever measured for collision-free motion planning.

Can you share a real-world example where your technology has been deployed?

Max Cao: One example is our work with Formic, a Robot-as-a-Service provider based in Chicago. They deploy a lot of palletizing solutions, and before using Jacobi, they programmed palletizers using traditional methods, which could take weeks on-site.

With our technology, they’ve been able to drastically reduce that time. For instance, a robot that previously took about a month to program and validate on-site was deployed end-to-end in just one day using our software. This included the entire programming, testing, and validation process. Moreover, since its deployment, that robot has also been reprogrammed remotely, demonstrating the flexibility and efficiency of our solution.

What’s next for Jacobi Robotics? How do you see the company evolving?

Max Cao: Jacobi Robotics is fundamentally a software company. Even though we started with a specific vertical—palletizing—we focus on the software side. We partner with machine builders, integrators, and other companies to bring a complete solution to the market.

Looking ahead, we’re focusing on two main areas: First, extending our foundational technology, including motion planning and the tools around it, like our visualization tool, Jacobi Studio. We want to make these tools accessible so others can build their own applications using our technology.

Second, we’re looking to showcase new capabilities. Now that we’ve tackled palletizing, we’re not going to repeat the same type of application. We might move into something like depalletizing with computer vision, integrating our software with 3D cameras from various providers. We see a lot of potential in applications that require real-time trajectory planning in semi-unstructured environments, intricate tasks like assembly, and multi-robot cells.

Any final thoughts on what excites you about the future of Jacobi Robotics?

Max Cao: What excites me the most is the potential to move the field of automation forward. Our goal is to make advanced robotics more accessible and easier to deploy across various industries. The enthusiasm I have for this work is rooted in the belief that we’re contributing to something that will have a significant impact on the future of manufacturing and beyond.

PS: Do you have any questions for Max? Reach out to me at [email protected] and I can pass it on.

Here are some interesting resources I found while writing this article:

• Jacobi co-founders’ paper titled “Deep learning can accelerate grasp-optimized motion planning” (Science)

• Jacobi’s Chief Scientist Ken Goldberg on “Why we don’t have better robots yet” (TED)

• Jacobi’s launch video of the Jacobi Palletizer (Jacobi on Vimeo)

• AI is poised to automate today’s most mundane tasks (MIT Tech Review)