BitRobot, a robotics network built on Solana, has released 2,000 hours of urban navigation data as open-source material, aiming to tackle one of the biggest obstacles in embodied artificial intelligence: obtaining large volumes of real-world interaction data.

The project is developing a decentralized marketplace for robotics datasets, using blockchain technology to track contributions and reward participants. The approach is intended as an alternative to centralized teleoperation facilities, where large teams of workers remotely operate robots to generate training data.

The newly released FrodoBots-2K dataset contains 2,000 hours of urban navigation data — far exceeding the roughly 60 hours available in earlier public datasets — giving AI researchers a substantially larger resource for training navigation systems.

From remotely controlled sidewalk robots to open AI data

BitRobot was originally established under the name FrodoBots, with early work focused on sidewalk robots controlled remotely by gamers. Participants operated the machines during scavenger hunt-style activities, generating data from real-world navigation scenarios.

That information has since been compiled into FrodoBots-2K and made publicly available. According to the report, teams associated with DeepMind, Meta, and the University of California, Berkeley have already used the dataset in developing navigation models.

The project holds that real-world data remains one of the most significant constraints on robotics development. Jonathan Victor, president of BitRobot, has argued that the industry's needs go beyond computing capacity and model architecture, because robots must learn from unpredictable environments and interactions rather than exclusively from controlled laboratory conditions. The challenge mirrors the trajectory of large language models, which advanced rapidly once internet-scale text corpora became available; embodied AI has lacked an equivalent source of physical interaction data, and open releases like FrodoBots-2K are one attempt to close that gap.

To address this challenge, BitRobot has established task-specific subnets designed to gather different categories of robotics data. These cover urban navigation as well as tasks involving physical dexterity, allowing the network to target specialized datasets instead of relying on a single centralized collection system.

Wearable cameras expand data collection

One of BitRobot's primary tools is RoboCap, a wearable device priced at about $1,000 and fitted with six cameras. The equipment records first-person video and hand movements, enabling data collection while people carry out ordinary tasks in settings such as bakeries and factories.

The company plans to distribute the devices among workers and other contributors to broaden the range of real-world scenarios represented in its datasets. Participants are expected to be rewarded based on the novelty and usefulness of the information they provide.

A scoring mechanism is designed to favor unusual or valuable situations over repetitive data, which could encourage contributors to capture scenarios unlikely to emerge from conventional, highly controlled data-collection operations.

Solana provides the blockchain infrastructure

Solana serves as BitRobot's underlying blockchain infrastructure, supporting contribution tracking, accounting, and payments. Its relatively low transaction costs are intended to make compensating participants across a geographically distributed network practical.

Victor has highlighted Solana's developer tools and established developer community as key factors behind the decision to build on the blockchain.

BitRobot also uses Access ID credentials to let contributors build robotics-related reputations. Participants can earn digital rewards known as Bolts for their contributions, creating a system designed to link participation, reputation, and compensation.

By combining open datasets, wearable data-collection hardware, and blockchain-based incentives, BitRobot aims to create a robotics network in which contributors supply valuable real-world data and share in the benefits it generates.

Data ownership could shape robotics competition

The project's development comes as embodied AI becomes an increasingly important research area, with companies and academic institutions seeking better ways to train robots for physical-world tasks. Major technology firms have invested heavily in robot learning programs, and access to scarce real-world datasets has become a competitive factor among them — one reason open releases such as FrodoBots-2K, which lower the barrier for smaller labs and independent researchers, draw attention across the field.

For Solana, BitRobot represents another potential application beyond financial transactions and speculative activity. SOL was trading at $104.60 as of September 7, 2026, down 0.66% over the previous 24 hours, though the token's market performance remains separate from the network's underlying technology.

BitRobot's longer-term prospects will depend on whether its decentralized model can produce higher-quality and more diverse datasets at a competitive cost compared with centralized alternatives. If successful, the network-driven model could give robotics developers broader access to real-world training data while creating new incentives for individuals and businesses to contribute previously difficult-to-collect information.