THE CRUNCH
Skild AI has launched its S1 robot foundation model, designed to learn previously unseen, long-horizon tasks from a single video demonstration. The model aims to help robots adapt to changing manufacturing environments without extensive reprogramming.
The S1 model is intended to address the challenge of robots struggling to keep up with tasks that change, layouts that shift, and new products arriving on manufacturing floors and in warehouses. By learning from a single video, the model is intended to reduce the need for significant reprogramming when environments change.
Skild AI describes the model as a 'physical AI' foundation model, suggesting it is designed to handle the complexities of the physical world, including the dynamics of moving objects and the variability of real-world tasks.