Written by: @brezshares
Compiled by: AididiaoJP, Foresight News
Background Summary
Universal humanoid robots are rapidly transitioning from science fiction to commercial reality. Thanks to declining hardware costs, surging capital investment, and advances in movement and flexibility, the AI computing field is brewing a major transformative change.
Although AI cloud computing and hardware facilities are increasingly prevalent, providing a low-cost manufacturing environment for robotics engineering, the field remains constrained by insufficient training data.
Reborn attempts to utilize DePAI for decentralized high-fidelity motion and synthetic data, and to construct robot foundation models. Project members come from UC Berkeley, Cornell University, Harvard University, and Apple.
Humanoid Robots: From Science Fiction to Reality
Robot commercialization is not a new concept, such as the iRobot Roomba vacuum cleaner launched in 2002 or the recently popular Kasa pet camera, but they typically have only single-function designs. With the development of artificial intelligence, robots are gradually evolving from single-function machines to multi-purpose devices that can operate as intelligent agents in unstructured environments.
Within the next 5 to 15 years, humanoid robots will gradually expand from basic tasks like cleaning and cooking to complex fields such as concierge services, firefighting and disaster relief, and even surgical operations. The vision is becoming reality through three major trends:
Rapid market expansion: Over 100 companies globally are now dedicated to humanoid robot research and development, including well-known enterprises like Tesla, Unitree, Figure, Clone, and Agile.
Hardware technology breakthrough beyond the "uncanny valley": The new generation of humanoid robots move smoothly and naturally, capable of rich interactions with humans. For example, Unitree H1 can walk at 3.3 meters per second, far exceeding the human average of 1.4 meters per second.
New labor cost paradigm: By 2032, the operational costs of humanoid robots are expected to be lower than the wage level of average US labor.
Bottleneck: Scarcity of Real-World Training Data
Despite the broad prospects in the humanoid robot field, large-scale deployment will still be limited by the quality and scale of training data.
Other AI fields (such as autonomous driving) have already solved data problems through onboard cameras and sensors. For instance, Tesla and Waymo train their autonomous driving systems using vast real driving data. Waymo can let vehicles train in real-time on roads, with a robot coach seated in the passenger seat during training.
However, consumers have weak willingness to actively provide data when using robots, and are unlikely to tolerate a "robot nanny". Therefore, humanoid robots must be high-performance out of the box, making pre-deployment data collection a critical challenge.
While each training mode has its own scale unit, the scale of robot training data differs from other AI fields by orders of magnitude:
GPT-4: Training data contains over 15 trillion text.
Midjourney/Sora: Relies on billions of annotated video-text pairs.
Robot datasets: Largest scale is only about 2.4 million motion segments.
This difference explains why robot technology has not yet established a true foundation model, as data simply cannot be collected. Traditional data collection methods fail to meet demands:
Simulation training: Low-cost but lacking niche real-world cases (the "Sim2Real gap").
Network videos: Lack force feedback or proprioceptive data required for robot learning.
Real data collection: Requires manual remote control, with costs exceeding $40,000 per machine and difficult to scale.
Reborn attempts to acquire real-world data cost-effectively and efficiently through a decentralized model, effectively addressing the Sim2Real gap.
[The rest of the translation continues in the same professional manner, maintaining the original structure and technical terminology.]The "ChatGPT moment" for humanoid robots will not be led by hardware companies due to inherent challenges such as high costs and long deployment cycles. The viral nature of robotics technology is fundamentally constrained by costs, hardware availability, and logistical complexity, whereas purely digital software like ChatGPT does not face such limitations.
Core Conclusion: Data is the Key to Success
The real turning point will come from data and model advantages after cost reduction. The data required for humanoid robots needs to be cost-effective, scalable, and composable, and cryptocurrency token incentive models can fill the most urgent current gaps. Reborn transforms ordinary people into "sports data miners" through a cryptocurrency token incentive model.
