Search papers, labs, and topics across Lattice.
This paper benchmarks 14 sensor combinations for imitation learning on a Unitree G1 humanoid robot performing manipulation tasks. They found that in data-limited regimes, a minimal active stereo-camera setup outperformed complex multi-sensor configurations, achieving 87.5% success in spatial generalization and 94.4% in structured manipulation. The study introduces an open-source Unified Ablation Framework using sensor masking on a comprehensive master dataset to evaluate sensor combinations.
More sensors, more problems: a simple active stereo camera setup beats out complex multi-sensor rigs for humanoid robot imitation learning when data is scarce.
The complexity of teaching humanoid robots new tasks is one of the major reasons hindering their widespread adoption in the industry. While Imitation Learning (IL), particularly Action Chunking with Transformers (ACT), enables rapid task acquisition, there is no consensus yet on the optimal sensory hardware required for manipulation tasks. This paper benchmarks 14 sensor combinations on the Unitree G1 humanoid robot equipped with three-finger hands for two manipulation tasks. We explicitly evaluate the integration of tactile and proprioceptive modalities alongside active vision. Our analysis demonstrates that strategic sensor selection can outperform complex configurations in data-limited regimes while reducing computational overhead. We develop an open-source Unified Ablation Framework that utilizes sensor masking on a comprehensive master dataset. Results indicate that additional modalities often degrade performance for IL with limited data. A minimal active stereo-camera setup outperformed complex multi-sensor configurations, achieving 87.5% success in a spatial generalization task and 94.4% in a structured manipulation task. Conversely, adding pressure sensors to this setup reduced success to 67.3% in the latter task due to a low signal-to-noise ratio. We conclude that in data-limited regimes, active vision offers a superior trade-off between robustness and complexity. While tactile modalities may require larger datasets to be effective, our findings validate that strategic sensor selection is critical for designing an efficient learning process.
