SpikeATac: A Multimodal Tactile Finger with Taxelized Dynamic Sensing for Dexterous Manipulation
Eric T. Chang, Peter Ballentine, Zhanpeng He, DoGon Kim, Kai Jiang, Hua Hsuan Liang, Joaquin Palacios, William Wang, Pedro Piacenza, Ioannis Kymissis, Matei Ciocarlie
AI summary
Problem
Existing tactile sensors struggle to combine high-resolution dynamic sensing with static pressure sensing in a compact finger, and integrating these rich, hard-to-simulate dynamic signals into modern learning-based manipulation pipelines remains challenging.
Approach
The authors designed SpikeATac, a multimodal robotic finger embedding a 16-taxel PVDF array for high-frequency dynamic contact detection alongside capacitive pads for static pressure, and integrated it into a learning framework combining imitation learning and on-robot reinforcement learning with tactile rewards.
Key results
- First multi-taxel PVDF array integrated into a complete robotic finger with complementary capacitive static sensing
- Ultra-fast, delicate grasping of fragile objects via high-speed PVDF contact detection
- Integration of raw dynamic tactile signals into RLHF and tactile-reward learning pipelines
- First in-hand manipulation of fragile objects on a multifingered robot hand
Why it matters
Provides a practical hardware and learning framework for robots to safely and dexterously handle fragile, deformable objects in real-world applications.
Abstract
In this work, we introduce SpikeATac, a multimodal tactile finger combining a taxelized and highly sensitive dynamic response (PVDF) with a static transduction method (capacitive) for multimodal touch sensing. Named for its ‘spiky’ response, SpikeATac’s 16-taxel PVDF film sampled at 4 kHz provides fast, sensitive dynamic signals to the very onset and breaking of contact. We characterize the sensitivity of the different modalities, and show that SpikeATac provides the ability to stop quickly and delicately when grasping fragile, deformable objects. Beyond parallel grasping, we show that SpikeATac can be used in a learning-based framework to achieve new capabilities on a dexterous multifingered robot hand. We use a learning recipe that combines reinforcement learning from human feedback with tactile-based rewards to fine-tune the behavior of a policy to modulate force. Our hardware platform and learning pipeline together enable a difficult dexterous and contact-rich task that has not previously been achieved: in- hand manipulation of fragile objects. Videos are available at roamlab.github.io/spikeatac.