The field of robotic manipulation and tactile perception is moving towards more advanced and nuanced approaches to interaction with complex environments. Researchers are exploring new methods for designing subspaces for reduced order modeling, allowing for more efficient and accurate simulations of dynamic scenes. Additionally, there is a focus on developing more effective tactile sensing systems, including multimodal sensor-integrated grippers and self-powered intrinsic static-dynamic pressure sensors. These advancements have the potential to enable more robust and adaptive control in a variety of applications, including robotic grasping and manipulation. Notable papers in this area include: MagicGripper, which presents a compact and versatile multimodal sensor-integrated gripper for contact-rich robotic manipulation. SAVOR, which proposes a novel approach for learning skill affordances for bite acquisition in robot-assisted feeding. DyTact, which introduces a markerless capture method for accurately capturing dynamic contact in hand-object manipulations.
Advances in Robotic Manipulation and Tactile Perception
Sources
Force-Dual Modes: Subspace Design from Stochastic Forces
Towards Tangible Immersion for Cobot Programming-by-Demonstration: Visual, Tactile and Haptic Interfaces for Mixed-Reality Cobot Automation in Semiconductor Manufacturing
MagicGripper: A Multimodal Sensor-Integrated Gripper for Contact-Rich Robotic Manipulation
System-integrated intrinsic static-dynamic pressure sensing enabled by charge excitation and 3D gradient engineering for autonomous robotic interaction
SAVOR: Skill Affordance Learning from Visuo-Haptic Perception for Robot-Assisted Bite Acquisition
Efficient Tactile Perception with Soft Electrical Impedance Tomography and Pre-trained Transformer
DyTact: Capturing Dynamic Contacts in Hand-Object Manipulation
Grounded Vision-Language Interpreter for Integrated Task and Motion Planning
Robustness-Aware Tool Selection and Manipulation Planning with Learned Energy-Informed Guidance
ActivePusher: Active Learning and Planning with Residual Physics for Nonprehensile Manipulation
Midplane based 3D single pass unbiased segment-to-segment contact interaction using penalty method