Maven founders, CFO Khalid Derbas and CEO Hamza Derbas with the company’s robot. | Source: Maven Robotics
Since it was founded in 2024, Maven Robotics has created four iterations of its mobile manipulator, raised $100 million, and gone from early pilots to a production contract with its first customer. It currently has robots running 16 hours a day, five days a week, with up to 99% uptime, according to Hamza Derbas, co-founder and CEO of Maven Robotics.
Maven has moved fast, raising Series A funding last month, but the Santa Clara, Calif.-based company started its journey by focusing on which problems it wanted to solve.
“Maven started with a small group of people, and we together started looking deeply at industrial labor, and realized that it’s facing a grim reality,” Derbas told The Robot Report. “The world needs more and more industrial output every day, but the workforce available to do this work is dwindling.”
Currently, there are over half a million unfilled manufacturing jobs just in the U.S., and market analysts expect this number to grow to millions over the next few years, Derbas said.
“At the same time, we noticed that the humanoid robotics companies, the ones with legs, were starting with the human form factor and trying to work backwards into a problem to be solved,” he added. “That’s when it became very clear to us that there’s a massive opportunity for the right kind of team, carrying all the battle scars of building hard technologies, to come tackle this problem with a completely different approach, specifically geared towards activating autonomous labor at scale in these environments.”
Use case informs morphology, says Derbas
Right now, humanoids dominate conversations in the robotics industry. But legs aren’t the only way to create a general-purpose robot, noted Derbas.
“Instead of starting from a robot or specific morphology or even a model, we started from the jobs to be done, down to the dust, the dirt, and the grease on the factory floor,” he recalled. “Then we work up to the robot, asking the question: What does our robot need to do? How far does it need to be able to reach?”
One downside of a robot with a wheeled base is its footprint. A big, heavy base creates a very stable robot, but it’s not realistic in all manufacturing or logistics facilities.
“We found a way to make the footprint super small so that it can fit through doors, really go anywhere, while maintaining very large dynamic reach,” Derbas said. “We can reach to the back of a 48-in. [121.9 cm] pallet or deep into a storage shelf up to 3 m [9.8 ft.] high while carrying 30 kg [66.1 lb.] payloads.”
The Maven System as a whole is more than just the robots, which can collaborate autonomously with human workers or in robot fleets. It also includes the ability to communicate with customers’ infrastructure, so it can exchange crucial information, as well as the Maven Network.
“[The Maven Network] is a global fabric of shared learning,” said Derbas. “If a robot makes a mistake somewhere in California, a robot in Japan can learn how to deal with that mistake the next day.”
Maven Robotics worked with customers early on
Early on, Derbas said Maven started working with a few partners to co-develop a general-purpose robot. These customers were willing to offer their deep operations insights to inform system design.
“We didn’t ask them to just come see our robots in our lab,” said Derbas. “We insisted that we go out to the field on the factory floor or in the warehouse to watch people do the work and see where the friction lies, and we zeroed in very fast on a couple of workflows that we thought we could master quickly and scale up fast.”
Maven’s first two workflows were mixed-case palletizing and tote handling. “These became the seed of our journey to generalization, because we started deploying these and creating value for our customers, it opened up a bigger conversation,” Derbas said. “What if our robots could acquire more and more skills faster and faster?”
Maven worked with its customers to create a roadmap of use cases and a progressive approach to rolling out automation. The company deploys its robots to automate a task and improves operations as it learns from that deployment. When the mobile manipulator masters the skill, the customer scales up that deployment. Then, Maven starts working on the next task.
“By doing this iteration, we earn customer trust to deploy fast and learn from the real world as opposed to from the lab, and we unlock this data flywheel so that we can train more and more general models to compound the intelligence and in the long run,” Derbas said.
Register now and help us celebrate 20 years of RoboBusiness!
The road to generalization
Looking ahead, Maven Robotics plans to tackle assembly and fabrication of durable goods. This would allow its robots to build cars, ships, planes, and even other robots in the future.
“Industrial work is not for the faint of heart,” said Derbas. “The bar for reliability is very high. There’s less demand on wanting a robot to be perfectly generalizable in its intelligence on Day 1. Generalization is a journey; reliability is not. It’s required from Day 1.”
To handle a variety of tasks, Maven equipped its robots with a modular end-of-arm interface so it can adapt different grippers to the job at hand. Derbas said the company landed on two kinds of grippers, a vacuum suction gripper, and a parallel gripper with two fingers.
“We started with off-the-shelf grippers, they didn’t work well,” he said. “We ended up designing our own gripper, then we iterated on that. We’ve landed now on an extremely capable and miniaturized vacuum gripper.”
The team experimented a lot when it came to multi-fingered tooling. Derbas said Maven looked at grippers with five, four, and three fingers with different sensing and actuation methods.
“We realized that the hand itself doesn’t need to have that much actuation or degrees of freedom, because if you leverage the fact that the hands are sitting on seven degree-of-freedom arms, and we have two seven degree-of-freedom arms, the extra articulation and maneuverability you get can help you simplify the gripper,” he said. “Combined with the right intelligence, you can accomplish a lot.”
Maven Robotics looks to scale up deployments
Maven Robotics is working to scale up its deployments and get its robots working for more hours. Derbas said the company plans to run its fleets 24 hours a day, seven days a week before the end of the year. The company is offering its robots as a service (RaaS).
“We went from zero to production contracts with long-term recurring revenue contracts in about two years,” Derbas said. “We’re going to have a couple hundred robots on production contracts over the next year, and then hopefully in the thousands by 2028 as we gear up for mass production.”
For mass production, Derbas said there are three major areas the company will focus on: streamlining how it manufactures the mechanical and structural components of the robot, working with suppliers to bring down the cost of actuators, and eliminating redundancy in the robot’s electronic systems.