ALT Experiment 2 scenarios; lower figures display initial configurations, while upper figures depict motion along the 10 cm paths. The path in Exp. 2.1 and Exp. 2.3 is to the left of the operator (along the robot's x-axis), while in Exp. 2.2, it is toward the operator (along the robot's y-axis). Manipulability ellipsoids and their principal radii are shown for operator arm initial configurations. To enhance detail, ellipsoids are presented at half their actual size.

ALT Schematic of the barometric tactile sensor design and data processing pipeline. (a) Exploded view of the modular sensor architecture, illustrating the separation of the PCB and rubber layer, and how the barometers experience the external force (b) System flowchart of the dual-model approach for sensor data interpretation: An MLP model for super-resolution contact localization, and a CNN-LSTM network for object orientation estimation.

ALT Three rows of four images stacked with various colored graphics an spaces.

ALT Illustration of graph-structured data-driven tomographic tactile sensing. On the sensor with intricate geometry, EIT-GNN decodes the voltage measurements into conductivity distribution, accounting for the heterogeneous mesh connectivity.

ALT (a) Comparison between ELSA and several powered ankles from the literature in terms of build height and weight. Light blue dots indicate that we had to estimate one or more variables. *Weight and/or size does not include electronics and/or batteries. (b) User with amputation wearing the prosthesis. (c) ELSA rendering highlighting its main components. (d) Exploded view showing the removable battery pack and interchangeable sole.

ALT Robot deployment feasibility study: Left: Real-time deployment of our Modular SNN in a small indoor environment on a CPU. First, the reference dataset is collected and the Modular SNN is trained offline. During inference, the robot moves through the environment, collecting images, and predicting their place labels in real time. The reference and query images are collected at different times of the day. The images were collected at 1 Hz with the robot moving at approximately 1 m/s. The blue line and points represent the reference path and images, while the green line represents the query path. Correct predictions are marked with green crosses, and incorrect predictions are marked with red crosses. The four samples of the query and predicted places show the original images and their preprocessed forms, which are used as input to our Modular SNN. Right: Proof-of-concept robot deployment testing platform. We used an AgileX Scout Mini [60] robot equipped with an Intel RealSense D435 camera.

ALT iCub robot in action during an interactive game demonstration. (a) Robot is approaching the correct card while back bending to dodge the collision. (b) Robot is keeping a safe distance from the opponent's detected hand.

ALT User manipulates the haptic device to apply force with the follower robot tool to the kidney mock-up.

ALT Concept photo showing that the compact design of DIGIT Pinki is sufficient for integration as the fingertip of a Psyonic Ability hand (holding a US penny for size comparison).

ALT Presented VS elbow and distinct applications in transhumeral prostheses. (a) Bionic limb on the left adopts the VS-Elbow AA to locate all components below the elbow joint, thus being suitable for distal transhumeral amputations. The prosthesis on the right employs the VS-Elbow D2 to achieve a sparser mass distribution, enhancing user comfort in more proximal transhumeral amputations. Panel (b) highlights the two implementations of the VS elbow: the VS-Elbow AA on the left and the VS-Elbow D2 on the right.

ALT Experimental setup for the human–robot cooperative piano playing, where the human is in charge of the upper voice while the robot is providing the lower voice with chord accompaniment.

ALT Typical tissue resection with forceps and scissors. (a) Dual-arm robotic system for lesion resection, where the forceps are responsible for tissue traction, while the scissors are responsible for tissue cutting (inset adapted from [18]). The grasping force Fg control and pulling force Fp control are crucial for efficient and safe tissue traction. (b) and (c) show tissue slide and break results from insufficient and excessive grasping forces Fg, respectively. (d) insufficient tissue pulling force Fp results in cutting surface overlapping and prevents the scissors from performing continuous cuts. (e) unexpected tissue split results from excessive pulling force Fp.