AcTExplore: Active Tactile Exploration of Unknown Objects

TL;DR

AcTExplore addresses the challenge of reconstructing unknown 3D objects through active tactile exploration. It introduces a PPO-based framework that uses a trajectory-based, temporally enriched state, a 6-DOF discrete action space (plus touch-recovery), and a multi-component reward that combines contact area with an intrinsic exploration bonus to drive comprehensive surface coverage. The approach is trained on primitive shapes and demonstrates strong zero-shot generalization to unseen YCB objects, achieving high 3D surface coverage (IoU) and accurate geometry (Chamfer-$L1$) within a limited number of tactile interactions, and it transfers to real-world hardware without further fine-tuning. This work advances tactile perception by enabling efficient surface exploration and reconstruction, with potential downstream benefits for grasping and manipulation in cluttered or occluded environments.

Abstract

Tactile exploration plays a crucial role in understanding object structures for fundamental robotics tasks such as grasping and manipulation. However, efficiently exploring such objects using tactile sensors is challenging, primarily due to the large-scale unknown environments and limited sensing coverage of these sensors. To this end, we present AcTExplore, an active tactile exploration method driven by reinforcement learning for object reconstruction at scales that automatically explores the object surfaces in a limited number of steps. Through sufficient exploration, our algorithm incrementally collects tactile data and reconstructs 3D shapes of the objects as well, which can serve as a representation for higher-level downstream tasks. Our method achieves an average of 95.97% IoU coverage on unseen YCB objects while just being trained on primitive shapes. Project Webpage: https://prg.cs.umd.edu/AcTExplore

Figures (11)

• Figure 1: Reconstruction of a hammer. (a) showcases the trajectory of the tactile sensor in 3D space. (b) depicts the intermediate tactile readings on the hammer's surface, with the color gradient representing the passage of time. Following thorough tactile exploration, we achieve a complete object reconstruction (c), highlighting the effectiveness of our active strategy in exploring the entire reachable surface.
• Figure 2: Overview. This figure illustrates the key steps and components of AcTExplore in a scenario where the sensor moves upward along the jar's edge. We employed Temporal Tactile Averaging for state representation $f$ (Sec. \ref{['method:staterep']}) to encode consecutive observations, enabling the perception of movement on the sensor vital for learning dexterous actions. We also incorporate an Upper Confidence Bound (UCB) exploration as a bonus to encourage effective exploration.
• Figure 3: Depth readings sliding over hammer (a): Sensor aligned with object's surface, receiving more depth information and moving stably. (b): Misaligned rotation increases the probability of losing contact in future steps.
• Figure 4: Training results. For each row: The [top] compares state representations using AMB for the reward function, while the [bottom] showcases different reward settings using TTA for state representation. Note that episodes terminate when the agent surpasses 90% IoU, reaches the horizon steps, or reaches the workspace boundaries.
• Figure 5: Qualitative results on unseen YCB objects with different state and reward settings. We obtain point cloud data from active tactile exploration on the object's surface. To generate a mesh from the collected point cloud, we apply Poisson surface reconstruction algorithm kazhdan2006poisson. Further experiments are provided in supplementary materials.