Existing manipulation data collection methods typically offer either accurate robot-state acquisition through kinematic-chain-based teleoperation or convenient, portable operation through vision-based approaches. We present DEX-PHD, a portable handheld device that combines these advantages for dexterous manipulation by integrating a dexterous robot hand, continuous per-finger control, and multimodal logging. The system records proprioceptive information, fingertip contact data, and wrist-camera video, while the operator directly controls the robot hand through joint-level mapping. Its untethered design enables data collection across diverse environments. We demonstrate DEX-PHD on contact-rich tasks requiring coordinated multi-finger manipulation and deploy the collected demonstrations through open-loop replay on a physical humanoid robot. The results show that open-loop replay preserves the main temporal patterns of finger motion and task-relevant contact. Therefore, DEX-PHD provides a practical platform for portable, multimodal dexterous manipulation data collection.
The gripper can submerge under water and grasp objects with passive mechanical components, a passive rotational joint and particle jamming effect pads. These parts enhance shear resistance during manipulation.
Development of full-surface tactile sensing in a dexterous hand.
Article: https://ece.illinois.edu/news/ece-associate-professor-awarded-nsf-convergence-accelerator-grant
Edelstro is a facade cleaning robot which cleans facades of high-rise buildings mounted on a gondola. Edelstro consists of a manipulator, modular cleaning devices with obstacle-overcoming ability, and contaminant detection module.
The manipulator operates with adaptive force control algorithm and moves with two directions, y-direction and yaw-direction.
The cleaning devices consist of squeegee, nozzles, roller brush, and suction parts. The cleaning parts can be equipped according to the environment of the building, and the cleaning can be implemented by dry or wet method. Each cleaning module has suspension so that the cleaning device can overcome window frames passively.
The contaminant detection module can detect contamination on the facade by AI algorithm that the robot can clean effectively whatever type of the contaminant is.
The Edelstro was exhibited at 2019 The Big 5 International Building & Construction Show and CES 2020, 2019 Korea Tech Show.
A two degree-of-freedom (DOF) parallel robotic manipulator for facade cleaning robots including tri-star wheels and series elastic actuators (SEAs) was developed and optimized. First, a cleaning device which is equipped at the end-effector was developed by downsizing for increasing cleanable area near window frames while maintaining cleaning methods. Secondly, a 2 DOF manipulator was developed. Tri-star wheels and SEAs help the robot overcome window frames passively. The manipulator was optimized for the safety of the gondola and the robot by minimizing the change of reaction force during overcoming the frames.
Adaptive mechanisms' performances for cleaning modules as end-effectors of robotic manipulator were compared by a simulation tool. Three adaptive mechanism models with tri-star wheels and performance indices were introduced. The obstacles' height were 15mm, 30mm, and 40mm.
FCR-M1 is a facade cleaning robot mounted on a gondola on high-rise buildings.
A cleaning device, which has squeegees, nozzles, roller brush, and suction parts, was optimized by Taguchi's method for robust design.
A two degree-of-freedom(DOF) parallel manipulator was developed. This manipulator can move along y-direction and yaw-direction. The manipulator was optimized for increasing dynamic manipulability ratio of minimum to maximum value, which is related with cleaning performance.
Dream Maker is a reconfigurable 2.5 dimensional voxelized injection mold machine. This undergraduate thesis project was worked with a group in a "Creative Design Engineering Project 3" course at Yonsei University. This machine was able to re-configure its mold as 2.5 D model data were transferred to machine. The models can be created on a PLAYGROUND CAD Tool. After the mold was customized into and operated as what users made on the program, liquid material can be poured.