糖心原创

This research project presents a design of soft building blocks that can be disassembled and reconfigured to build different modular configurations of soft robots such as robotic fingers and continuum robots. To validate the approach, 2D and 3D case studies of bio-inspired designs are demonstrated: first, soft fingers are introduced as a case study for grasping complex and delicate objects. Second, an elephant trunk is used for grasping a flower. Third, a walking legged robot. These case studies prove that the proposed modular building approach makes it easier to build and reconfigure different types of soft robots with multiple complex shapes.

Flexible electromagnetic linear actuators (FLAs) provide a modular, 3D-printable solution to the limitations of traditional rigid, linear systems, enabling actuation in nonstraight configurations. A novel, computationally efficient mathematical model based on noncoaxial coil decomposition is presented to predict output force, validated experimentally across various 3D bending angles.

 

This work presents a macro–mini collaborative manipulator system for performing welding in confined industrial environments. A 6‑DoF macro robotic arm positions a novel 2‑DoF slim mini‑manipulator inside narrow spaces, where the mini‑manipulator’s serial mechanism provides precise end‑effector (torch) orientation. Dedicated kinematic and Jacobian models enable accurate coordinated control, including compensatory motions between the two manipulators to correct end‑effector errors. Validation experiments in constrained scenarios demonstrate reliable access, pose control, and welding functionality. The system offers a practical solution for in‑situ welding tasks where conventional robots cannot physically reach.

Robotic welding systems are essential in modern manufacturing but remain limited by the lack of embedded human expertise. Traditional systems require extensive manual programming for each new task, restricting efficiency in high-mix, low-volume scenarios. This project bridges the gap between human welding skills and robotic execution by developing a systematic framework that transforms implicit human expertise into actionable digital knowledge for automation.

This project investigates solutions in the areas of force sensing, force/position control, and haptic control for continuum robots. To navigate narrow and constrained environments, continuum robots possess unique characteristics distinct from traditional rigid robots, including high flexibility, a small diameter-to-length ratio, and low stiffness. These structural and operational features introduce significant challenges in modelling, sensing, and control.

This project develops a smart machining ecosystem for thin‑wall structures, integrating digital twins, adaptive automation, and human‑cyber‑physical systems. A fast mass‑spring‑lattice digital twin predicts machining‑ and clamping‑induced deformations in real time, enabling digital‑twin‑assisted corrective toolpaths. In the roughing habitat, cutting forces feed the twin to generate immediate compensation paths. In the finishing habitat, a sensorized intelligent fixture provides closed‑loop clamping control using either a fully automated robotic clamping system or a human‑assisted mobile‑app interface. By combining multi‑stage machining, ID‑linked toolpaths, and deformation‑aware corrections, the ecosystem improves geometric accuracy and demonstrates a scalable framework for smart, interconnected machining environments.

This project develops fully 3D-printed capacitive sensors in which the sensing function is built directly into the structure rather than attached afterwards. Using dual-material FDM, carbon-based conductive PLA and insulating PLA are printed together to form a compliant beam–mass structure with embedded electrodes. The printed geometry is designed to convert local vibration, acceleration, and deformation into measurable capacitance changes, while a dual-channel differential configuration and algorithm approach help reduce common environmental disturbances.

Through iterative design, modelling, fabrication trials, and controlled testing, the work examines both the potential and limitations of structure-led sensing, including resonance behaviour, baseline drift, manufacturing tolerance, and signal reliability. The printed sensing structures are also explored in a milling-fixture environment to understand how process-related vibration and deformation can be transmitted and identified. The project contributes a practical route towards low-cost, customisable, structurally integrated sensing for monitoring and smart manufacturing applications.

Human skill-based robotic control to perform critical manufacturing operations (e.g., repair and inspection for high-value assets) can reduce scrap rates and increase overall profitability in the industrial community. In this project, a human–robotic collaborative control system is developed for accurate path tracking subject to unknown external disturbances and multiple physical constraints. This is achieved by designing a model predictive control with a sliding-mode disturbance rejection term. To rule out the possibility of the constraints violation caused by external disturbances, tightened constraints are formulated to generate the control input signal. The proposed controller drives the robotic system remotely with enhanced smoothness and real-time human modification on the outputted performance so that the human experience can be fully transferred to robotic systems. The efficacy of the proposed collaborative control system is verified by both Monte–Carlo simulation with 200 cases and experimental results including tungsten inert gas welding based on a universal robot 5e with 6 degree-of-freedom.