91视频

Robotics researchers often spend weeks, or even months, simply getting a new robot up and running before they can begin testing new behaviors. Researchers in the 91视频 have developed an open-source software framework designed to eliminate much of that setup work, making it easier to deploy AI systems across different robots without rebuilding software from scratch.

The framework, called (RIO), provides a unified interface for robot control, data collection, teleoperation and AI deployment. By allowing researchers to move more easily between robotic arms, humanoids and other platforms, RIO addresses one of the most persistent challenges in robotics: the extensive engineering work required before research can begin.

鈥淭he biggest bottleneck in robot learning research isn鈥檛 ideas, it is infrastructure,鈥� said , an associate research professor in the (RI). 鈥淪tudents can spend an entire semester or even their entire first year simply setting up a robot before they can begin their research. RIO gives researchers and engineers a lightweight, modular foundation for deploying robots quickly on any platform.鈥�

The framework鈥檚 emphasis on accessibility became apparent during testing. Reya Shukla, an undergraduate intern with machine learning experience but no robotics background, was asked to unpack a robotic arm and configure it for teleoperation using RIO. Following the framework鈥檚 documentation, she went from opening the box to controlling the robot in about two hours.

That ease of setup reflects the problem RIO was designed to solve. While advances in robot intelligence have accelerated in recent years, the infrastructure needed to support those systems has lagged. Research groups often build custom software for individual robots, making it difficult to share tools, data and AI models across platforms. Much of the code developed for one robot must be rewritten when moving to another system, making it difficult to reproduce results or share advances across labs.

RIO is built around modular software components that can be reused across different robots and research projects. Instead of rebuilding the same infrastructure each time they switch hardware, researchers can combine existing components and customize the system to fit their needs. The approach is intended to make robotics development more accessible while improving consistency and reproducibility across experiments.

According to RI Ph.D. student , RIO arrives at a pivotal moment as new AI approaches are changing how robots are controlled. Existing robotics software was largely designed before these more general-purpose intelligence systems became common, creating a need for infrastructure that can support them.

鈥淭oday, everyone talks about needing more data,鈥� said , another RI Ph.D. student on the RIO team. 鈥淏ut robot data doesn鈥檛 come from thin air. You need robots to collect it, and that takes robot infrastructure. We鈥檝e been missing some shared building blocks that other AI fields have, the ones that let researchers build on each other鈥檚 work instead of starting from scratch every time.鈥�

Vernon Luk, an incoming student in the program and Megan Lee, a recent graduate of the (MRSD) program, found that RIO also simplified working with multiple robots while developing policies for humanoid and bimanual platforms.

鈥淪ince the building blocks are designed to be swappable, you can use the same pipeline for all the robots you work with,鈥� Luk said. 鈥淵ou don鈥檛 have to write special code to collect data or train policies for different platforms, even when they have additional cameras or robot arms.鈥�

By providing a shared foundation for robot control, data collection and AI deployment, the framework could help accelerate research across the field while making robotics more approachable for newcomers.

鈥淚n industrial robotics settings, this kind of infrastructure really matters because real deployments rarely involve just one robot, one sensor setup or one fixed environment,鈥� said , lead research scientist at the and courtesy faculty in the RI. 鈥淩IO helps make robot learning systems more reusable across platforms, which can shorten the path from a research prototype to something that can be tested and adapted in the real world.鈥�

While RIO is still an active research project, some team members are continuing to build on the technology through , a startup co-founded by Oh that focuses on simplifying robot deployment and accelerating robot learning. Future work will focus on expanding hardware support and further lowering the barrier to bringing new robots online, with a longer-term vision of building robotics foundation models that enable robots to rapidly and autonomously adapt to new tasks and environments across platforms.

In addition to Oh, Ortega-Kral, Xing, Luk, Lee and Francis, the RIO team includes RI Assistant Professor Guanya Shi; RI Ph.D. student Arthur Fender Coelho Bucker; incoming MSR student Vernon Luk; RI visiting scholar and Delft University master鈥檚 student Junseo Kim; biomedical engineering master鈥檚 alumnus Owen Kwon; MSR student Angchen Xie; MRSD students and alumni Yifu Yuan, Deepam Ameria and Bhaswanth Ayapilla; electrical and computer engineering graduate student Nikhil Sobanbabu; and Jaycie Bussell, an engineer at Bosch Manufacturing.听