Creepy Crawlers: General AI Robots Take a Shocking Step Forward
Artificial Intelligence
Robotics
AI Adaptation
Machine Learning
Skild AI
Robot Brain
General AI
8
Adaptive Intelligence
Media Hype
7/10
Real Impact
8/10
What is the Viqus Verdict?
We evaluate each news story based on its real impact versus its media hype to offer a clear and objective perspective.
AI Analysis:
While the 'creepy' nature of the adaptability certainly generates media buzz, the underlying technology—a truly generalist AI robot—has the potential for a massive, long-term impact, far exceeding current social media trends.
Article Summary
Skild AI’s development of ‘Skild Brain,’ a generalist AI algorithm trained on a diverse range of robotic hardware and tasks, represents a significant leap in robotic AI. Unlike traditional training methods relying on teleoperation or simulation, Skild’s approach uses a single algorithm to control numerous robots across varied environments. This leads to a remarkably adaptable AI capable of responding to unforeseen circumstances, such as losing limbs or experiencing morphological changes. The core innovation lies in the robot’s ability to ‘learn’ and apply its knowledge to entirely new situations, mimicking the adaptive capabilities seen in large language models. The ‘LocoFormer’ model, a smaller version, demonstrated impressive adaptability, controlling different robot shapes and even walking on its hind legs after its legs were removed. The system’s online learning capability, coupled with aggressive domain randomization, further enhances its robustness. This technology is attracting significant attention, with competitors like the Toyota Research Institute and Physical Intelligence also pursuing similar generalist AI models. Skild’s success, bolstered by a recent $300 million funding round, suggests a future where robots can seamlessly transition between tasks and environments, potentially revolutionizing industries from manufacturing to logistics.Key Points
- Skild AI developed ‘Skild Brain,’ a generalist AI algorithm trained on a diverse range of robotic hardware.
- The ‘LocoFormer’ model demonstrates impressive adaptability, able to control different robot shapes and walk on its hind legs after losing its limbs.
- The system’s online learning capability and aggressive domain randomization contribute to its robustness and ability to adapt to unforeseen circumstances.