Spark-0: Scaling Spatially Aligned Tactile Pretraining for Robot Dexterous Manipulation
Spark-0 is a vision-tactile-language-action (VTLA) model that uses touch in two ways: as training supervision for visual tokens and as input to condition the action expert. By pretraining on 1,000 hours of human tactile data, the model achieves state-of-the-art performance in dexterous manipulation benchmarks, showing that touch is a scalable way to improve vision-only policies.