IROS 2026: Viam to showcase box-opening robot. Viam plans to demonstrate BoxBot, a robotic arm designed to open taped cardboard boxes, at the IROS 2026 robotics conference in Pittsburgh. BoxBot is an ongoing research project that combines image-based visual servoing with a vision-language-action model. The visual servoing system uses camera feedback to guide a cutting tool along tape seams, which Viam said are typically 1 to 2 millimeters wide. A fine-tuned SmolVLA model handles the subsequent task of opening and adjusting the box's four flaps. The two systems operate as services on the same machine. The VLA model performs inference locally on an Nvidia Jetson Orin Nano using Viam's machine-learning model service. The robot also uses conventional motion planning for other parts of the application. Viam said two of its engineers collected 125 demonstrations for training the system over about five hours across four days. Using a virtual-reality controller, the engineers generated demonstrations while Viam's software synchronized robotic-arm data with feeds from two cameras. The company then fine-tuned SmolVLA using LeRobot on Hugging Face Jobs and deployed the resulting policy to the robot. Viam said the project did not require a custom model architecture. The company has made the BoxBot code and dataset publicly available on Hugging Face for researchers seeking to reproduce or benchmark the system. At IROS, Viam plans to operate BoxBot alongside two additional demonstration stations. One will allow attendees to control a robotic arm using the VR setup employed to generate BoxBot's training data. Another will use an SO-101 leader-follower configuration to demonstrate arm mirroring, episode recording and replay, and synchronized data capture. Shannon Bradshaw, Viam's senior vice president of product and education, said the project was intended to combine learned policies with methods such as visual servoing and classical motion planning within a single robotic application. Viam's platform provides hardware abstraction, computer vision, data capture, visual servoing and edge machine-learning deployment through a common set of APIs. Developers can configure hardware using JSON and write application logic in Python, Go or TypeScript. The platform can also operate alongside ROS 2-based robotic systems through Viam's ROS 2 integration module.

Viam debuts box-opening robot at IROS 2026. Trained on 125 teleoperated demonstrations collected in about five hours, BoxBot pairs image-based visual servoing with an open source VLA, no custom architecture required, on one data path built on Viam. Viam, the software platform for robotics, will debut BoxBot at IROS 2026, one of the world's leading robotics research conferences held in Pittsburgh from September 27 through October 3. BoxBot is an ongoing Viam research project: a robot arm that opens taped cardboard boxes using image-based visual servoing to cut the tape and a vision-language-action (VLA) model, trained on human demonstrations, to open the flaps. More headlines. Articles. BoxBot specifies what it can and learns what it can't: Visual Servoing: No two sealed boxes are alike, so the robot uses image feedback to guide the cutting tip to tape seams that are typically 1 to 2mm wide. Learned Policy (VLA): A fine-tuned SmolVLA model handles the deformable, contact-rich task of peeling open and adjusting four box flaps. Both systems run as native services on the same machine, with the VLA policy inferring locally on an NVIDIA Jetson Orin Nano via Viam's ML model service. Less time on data plumbing, more time on research Researchers often spend weeks building teleop rigs, syncing cameras to joint states, and formatting episodes for training, then rebuild it all for the next arm. On Viam, that capture layer already exists, making it easy to add an input device as a module, record synced episodes, and copy a config fragment to stand up another station. Two Viam engineers collected BoxBot's 125 demonstrations in about five hours over four days using a VR controller. Viam synced the arm data and two camera streams to the cloud, the team fine-tuned SmolVLA with LeRobot on Hugging Face Jobs, and deployed the policy back to the robot. No custom model architecture, no task-specific engineering: the base model and 125 episodes were enough for a working policy. The code and dataset are open on Hugging Face at https://huggingface.co/viamrobotics, so researchers can train on them, benchmark against BoxBot, or reproduce the setup. See BoxBot in development at Viam's NYC robotics lab, here. At the booth IROS attendees can see BoxBot run, drive the arm themselves, and record data on a separate station: BoxBot: Watch the arm cut the tape seam and open the flaps. VR teleoperation station: Control the arm using the same VR setup that produced BoxBot's training data. SO-101 leader-follower station: Move a leader arm, watch the follower mirror it, record and replay an episode, and see the synced data in the Viam app. "The question isn't whether to use a VLA, it's where," said Shannon Bradshaw, Ph.D., SVP of Product and Education at Viam. "Learned policies are the right tool for contact-rich manipulation tasks that are hard to model analytically, and methods like visual servoing enable precise solutions in dynamic environments. This robot also uses classical motion planning for other elements of the application. With many frameworks, this would require extensive ad hoc scripting to bring everything together in one application. The Viam platform provides an integrated solution that enables us to combine the hardware, software, and models this robot uses in a coherent solution we can easily iterate on, updating components as needed and adding new ones as we build new capabilities. Our implementation is publicly available, so robotics and physical AI researchers can check our work or build on it." Viam, the software platform for robotics, runs BoxBot's hardware abstraction, visual servoing, computer vision, data capture, and edge ML deployment as services behind one set of APIs. Developers connect hardware in JSON, write logic in Python, Go, or TypeScript, and pull modules from the Viam Registry. ROS 2 teams can run Viam alongside existing robots through the viam-ros2-integration module. IROS attendees can visit Viam at Booth 828. To book a technical demonstration or meeting with the Viam engineering team, reach out to Viam's SVP of Product and Education, Shannon Bradshaw, at [email protected]. About Viam Viam is the software platform for building, deploying, and managing robotics applications. With Viam, engineers, startups, and large enterprises can develop for the physical world the way they develop anything else. Viam works with teams across manufacturing, marine, climate, logistics, entertainment, and more. Founded in 2020 by MongoDB co-founder and former CTO Eliot Horowitz, Viam has its headquarters and robotics lab in New York City. viam.com

Viam is showcasing BoxBot, a box-opening robot, at IROS 2026 in Pittsburgh from 27 September to 3 October. The robot arm uses image-based visual servoing to cut tape and a vision-language-action model to open flaps. Two Viam engineers collected 125 demonstrations in about five hours over four days using a VR controller. The team fine-tuned SmolVLA with LeRobot on Hugging Face Jobs and deployed the policy back to the robot. The code and dataset are available on Hugging Face. BoxBot combines visual servoing for precise tape cutting with a learned policy for handling deformable box flaps, running both systems on an NVIDIA Jetson Orin Nano. Conference attendees can see BoxBot operate and try VR teleoperation at Booth 828.