In a warehouse or factory, a robot can move materials while people make decisions and handle work that calls for human judgment. Robust AI built Carter around that division of labor: a mobile robot designed to operate alongside workers, with controls that let them guide it when needed.
A robot workers can steer
Carter resembles a wheeled utility cart. It has a force-sensitive handlebar, an AI-powered vision system and holonomic wheels that let it move in any direction. Workers can push or pull it with one hand, gesture to direct it, or use its onboard touch screen.
The design gives workers ways to interact with the robot directly as conditions change. They can also cut across its path. That flexibility matters in workplaces where people and machines share space and work does not always follow a fixed route.
Robust AI’s robots are deployed in logistics warehouses and factory settings. They can help workers pick items, stock shelves and move materials between locations autonomously. Customers can introduce the robots in phases and choose workflows that suit their operations.
Automation that can fit existing operations
One workflow has a robot guide a worker to locations where items need to be retrieved. Another has it follow a worker to loading areas. In the main material-moving workflows, Carter can carry items across large facilities, reducing the need for a person to make each trip themselves.
Rodney Brooks, Robust AI’s chief technology officer, described a worker loading a robot and then asking it to take the load to another station. The station might be 400 feet away. In that case, the worker can say “done,” and the robot makes the trip.
The company chose logistics and manufacturing after years of talking with potential customers. Anthony Jules, Robust AI’s chief innovation officer, argues that many factories are not set up for large, costly automation systems. The company’s robots can be added as needed, rather than requiring customers to rebuild an entire operation around automation.
That approach treats automation as a way to support existing work. A robot can take on transport or help with item retrieval, while people continue to guide the work and respond to the circumstances around them.
Why a robot must understand the work
Robust AI’s focus on worker interaction reflects a practical problem: a robot can disrupt a workplace if it does not account for what people need to do. Leila Takayama, the company’s vice president of human robot interaction, pointed to a peer-reviewed study of robots used in hospitals to gather dishes, move sheets and transport equipment.
In many hospitals, researchers found robots turned off and against the walls. Nurses said the robots broke. Brooks described a different underlying issue: when nurses rushed with a patient on a gurney, robots stopped in the hallway and blocked their way. The machines were interfering with the nurses’ primary tasks.
The example helps explain the company’s development approach. Robust AI says it studies workers’ main tasks and builds features intended to speed those tasks up, whether a person or a robot performs them. A machine that simply stops whenever it detects people may avoid a collision, but still make the workplace harder to navigate.
Building a company around collaboration
Jules and Brooks co-founded Robust AI. Jules first met Brooks as a graduate student at CSAIL in the 1990s. Their labs were on the same floor, and they stayed in touch after Jules graduated. In 2019, they met again at a robotics conference, where Brooks raised the idea of a robotics software company he compared to a video game engine for training robots. Robust AI was founded later that year.
The company’s plans shifted during the Covid-19 pandemic, when it turned toward robots that could disinfect surfaces. After the founders learned that Covid-19 was not transmitted via surfaces, they had to reconsider how to use the technology they had been developing to help robots understand the world. Their next direction meant building hardware as well as software.
Carter reflects that turn toward a physical robot intended to work with people in real facilities. The company’s central premise is that workplace performance depends on the combined efforts of robots, people and backend systems. Giving workers a practical way to direct a robot is part of making that combination work.