Investigating Active Sampling for Hardness Classification with Vision-Based Tactile Sensors
This work addresses sample-efficient hardness perception for robotics, though it appears incremental as it applies existing active sampling methods to a specific tactile sensing task.
The paper tackles the problem of sample-efficient hardness classification using vision-based tactile sensors by investigating information-theoretic active sampling strategies. The results show that active sampling approaches achieve 88.78% average accuracy, surpassing both random sampling (baseline) and human testers (48.00%).
One of the most important object properties that humans and robots perceive through touch is hardness. This paper investigates information-theoretic active sampling strategies for sample-efficient hardness classification with vision-based tactile sensors. We evaluate three probabilistic classifier models and two model-uncertainty-based sampling strategies on a robotic setup as well as on a previously published dataset of samples collected by human testers. Our findings indicate that the active sampling approaches, driven by uncertainty metrics, surpass a random sampling baseline in terms of accuracy and stability. Additionally, while in our human study, the participants achieve an average accuracy of 48.00%, our best approach achieves an average accuracy of 88.78% on the same set of objects, demonstrating the effectiveness of vision-based tactile sensors for object hardness classification.