AI is Far More Than Just LLMs! In his latest ultimate AI plan, Musk boldly stated that 80% of Tesla's future value will come from robots! As robotics concept stocks surge wave after wave, Web3 robot projects are also accelerating, igniting a new wave of enthusiasm. How will a decentralized, on-chain collaborative intelligent ecosystem reshape our future?
Follow Biteye to unlock the rising stars in the Web3 robotics track!
-
@openmind_agi
Keywords: Operating System, Decentralized Coordination Layer
OpenMind raised $20 million, hoping to enhance robots' understanding and learning capabilities through an open-source system. Its main products include:
OM 1: An AI-native operating system, dubbed the "Android for robots," hardware-agnostic and open-source. It enables perception, memory, planning, and control capabilities, helping robots understand the world and take relevant actions based on context. Previously, a hackathon participant used OM 1 to build an Alzheimer's care robot within 48 hours.
FABRIC: A decentralized coordination layer. Using blockchain, it allows robots to share data, learn quickly (e.g., language), and supports consensus, execution, and settlement.
-
@peaq
Keywords: Layer 1, Machine Economy
peaq raised $21 million and is a Layer 1 blockchain focused on DePIN and the Machine Economy.
peaq enables robots, drones, sensors, and other machines to obtain self-sovereign IDs on the blockchain, conduct peer-to-peer payments, store data, and perform automated operations. It enables use cases like robot rentals, drone deliveries, and autonomous machine collaboration, building a complete "Machine Economy" where devices like robots can earn money and interact independently, much like AI Agents.
Built on the Substrate framework, it's compatible with Polkadot and connected to over 90 blockchains. Currently, there are over 50 DePIN projects on the peaq chain, including Silencio (noise monitoring), MapMetrics (drive-to-earn), Farmsent (agricultural supply chain), and Natix (smart driving). The network connects over 4.5 million devices and machines, providing decentralized network services for more than 20 industries including energy, mobility, transportation, agriculture, and environment.
-
@GEODNET_
Keywords: Real-time Positioning Network
GEODNET raised $15 million, focusing on positioning services—building the world's largest decentralized Real-Time Kinematic (RTK) positioning network to provide centimeter-level accuracy Global Navigation Satellite System (GNSS) services.
The project was initially built on Polygon and has now expanded to Solana.
The network allows anyone to purchase and install GEODNET's hardware miner devices ("satellite miners") to collect and upload RTK correction data to the network, earn token rewards, and ultimately empower the network to provide high-precision positioning support for applications like robots, autonomous vehicles, drones, and the metaverse.
The network currently covers 145 countries with over 19,000 stations and has delivered over 207,178 GB of RTK data.
-
@psdnai
Keywords: Physical Data Collection
Poseidon, incubated by @StoryProtocol, raised $15 million in seed funding. It primarily addresses the shortage of training data at the physical level for applications like robots and multimodal AI Agents.
Poseidon is building a decentralized data layer, based on Story, to collect, organize, and manage real-world data (such as video, audio, images, location, etc.), ultimately providing IP-protected training data to accelerate AI model innovation.
Currently, an app is live, starting with sound information, allowing users to upload audio information in the app to train AI models, enabling them to handle speech in different accents, dialects, and environments.
-
@PrismaXai
Keywords: Teleoperation Platform (Tele-op), Training Data
PrismaX, led by a16z CSX, raised a total of $11 million. It aims to use human intervention to help robots ultimately achieve highly autonomous capabilities.
It currently offers a teleoperation (tele-op) platform that allows users to remotely control robotic arms via a browser. These operation data are converted into high-quality training data to optimize models. Operators are rewarded with platform tokens in the process.
The platform plans to remotely connect to various well-known robots like Unitree and Ubtech in the future.
Short-term focus: Teleoperation and visual data for model training.
Medium-term: Enabling robots to perform practical tasks through remote human guidance.
Long-term: Ultimately achieving full or high autonomy for robots.
-
@NRNAgents
Keywords: Teleoperation Platform (Tele-op), Reinforcement Learning, AI Agent
NRN Agents is a decentralized AI Agent development and training platform running on Arbitrum. Starting with AI Agents in games, it is gradually expanding its business to the AI Agents behind embodied robots—providing efficient training and deployment solutions for robot AI Agents through a "Sim-to-Real" framework.
Users can directly control simulated robots via a browser. Each operation generates high-quality behavioral data, providing training materials for imitation learning in robot systems. Simultaneously, the AI Agents in the system can continuously optimize through reinforcement learning based on real interactions without requiring complete retraining.
The project is in its early stages, primarily focusing on robotic arm (RME-1) operation. It will later expand to robots' competitive sports abilities like combat and track and field.
-
@AukiNetwork
Keywords: DePIN Network, Physical World Data, Location Data
Auki Network wants to build a decentralized neural network for robots and smart machines to give them a better understanding of the physical world. Its core business is building a DePIN network that allows various devices (like robots, smart glasses) to share location and perception data in real-time, forming collaborative spatial understanding.
The network is based on the Posemesh protocol, including compute nodes providing computational resources, motion nodes (smart devices like robots providing motion and sensor data), reconstruction nodes using the former to generate 3D maps or spatial models, and domain nodes managing these 3D spaces. These nodes are rewarded with tokens based on their contributions.
-
@RoboStack_io
Keywords: Abstract Hardware Complexity, Simplify Hardware-Software Integration
RoboStack is a cloud-native simulation and integration platform. It not only provides realistic environment modeling and scalable computing infrastructure for robot development and deployment but also, through the RCP (Robot Context Protocol), establishes a standardized communication protocol between different hardware, software, and communication standards, simplifying the integration of underlying robot hardware with software-layer AI Agents.
The ROBOT Coin has been launched on the virtuals platform. Users can stake on virtuals and vote for their favored robot projects in SIM lord for a chance to receive rewards.
-
@frodobots
Keywords: Sidewalk Robot
FrodoBots Lab is a robotics lab promoting real-world data collection and embodied intelligence development through gamification. Its core product, Earth Rover, is a remotely controllable sidewalk robot (a robot that walks on sidewalks) priced at $249-399, usable for gaming, data collection, and AI research. FrodoBots has deployed hundreds of Earth Rovers globally and open-sourced a 2000-hour real-world driving dataset.
Through innovative robot hardware, decentralized platforms, and AI frameworks, FrodoBots has built an ecosystem blending entertainment and research. Its main projects include:
-
@BitRobotNetwork
Keywords: Decentralized Subnet Architecture
BitRobot Network is a Solana-based decentralized platform jointly developed by FrodoBots Lab and Protocol Labs, with $6 million in funding. The network uses a subnet architecture where each subnet contributes specific resources like computing power, robot fleets, datasets (real or synthetic), or AI models, covering various forms from sidewalk robots to humanoid robots. The first subnet, ET Fuji, is live and detailed below.
-
@ET_Fugi
Keywords: Game, Teleoperation Platform (Tele-op)
ET Fugi (Extra-Terrestrial Fugitives) is a real-world robot game where players remotely control Earth Rovers to capture "alien" NFTs, simultaneously generating valuable sidewalk data for robot AI training and earning rewards like FrodoBots Points (FBP).
-
@samismoving
Keywords: AI Agent, Framework
SAM can collect operational data through Earth Rovers to enhance overall AI Agent capabilities. It also live-streams parts of its adventures on X (Twitter), showcasing specific interactions, competitions, or data collection scenarios, offering both entertainment and educational value.
SAM was initially an AI Agent co-developed by FrodoBots and @virtuals_io for remotely operating Earth Rover robots. Later, it evolved into the core framework for launching AI Agents on the Robots.Fun (Launchpad) platform. All subsequent AI Agents will be built based on SAM's architecture and pay it a 5% fee share.
-
@robotsdotfun
Keywords: AI Agent Launch Platform
Robots.Fun is a launch platform for robot AI Agents. It collects robot data through daily competitions (like capturing alien NFTs) to enhance AI Agent capabilities. To create an AI Agent on Robots.Fun, one must own or purchase an Earth Rover robot to serve as the AI Agent's "body."
-
@UFBots
Keywords: Robot Combat, Teleoperation Platform (Tele-op)
Ultimate Fighting Bots is a free robot fighting club where users can remotely control Booster T1-based humanoid robots for combat. The platform generates real-world data for training embodied AI while providing users with an immersive entertainment experience.
-
Summary
After reviewing these projects, we can clearly see two core advantages of the Web3 robotics track: -
The decentralized advantage of blockchain has the potential to break down the scarcity barriers of robot models and data.
-
On-chain ownership verification and incentive mechanisms inject new vitality and sustainability into the robot economy.
