Synthetic, degradable macromonomers have been developed to serve as ink for 3D printing technologies based on direct-ink-writing. The macromonomers are purposely designed to be cross-linkable under the radical mechanism, to impart hydrophilicity to the final material, and to have rheological properties matching the printer's requirements. The suitable viscosity enables the ink to be printed at room temperature, in absence of organic solvents, and to be cross-linked to manufacture soft 3D scaffolds that show no indirect cytotoxicity and have a hydration capacity of up to 100% their mass and a compressive modulus in the range of 0.4-2 MPa.

4D printing technology, as a new generation of Additive Manufacturing methods, enables printed objectsto further change their shapes or other properties upon external stimuli. One main category of 4D printingresearch is 4D printed thermal Shape Memory Polymer (SMP). Its morphing process has large time delay, isnonlinear time variant, and susceptible to unpredictable disturbances. Reaching an arbitrary position with highprecision is an active research question. This paper applies the Reinforcement Learning (RL) method to developan optimal control method to perform closed loop control of the SMP actuation. Precise and prompt shapemorphing is achieved compared with previous control methods using a PI controller. The training efforts of RLare further reduced by simplifying the optimal control policy using the structural property of the prior trainedresults. Customized protective visors against COVID-19 are fabricated using the proposed control method.

Four-dimensional printing (4DP) is a newly emerged technology that uses smart materials for additive manufacturing and thus enables shape and/or property change upon stimulus after the printing process. Present study on 4DP has been focused on open loop stimulus, which can hardly ensure high shape precision and predictable final states. In this paper, a new closed loop 4DP (CL4DP) process supplementing 4D printed actuation with closed loop control methods is proposed. Image feedback is used for enhancing the conventional open loop 4DP morphing process and a controller is implemented to regulate the intensity of the stimulus accordingly in real-time. To achieve precise control, a nonlinear affine system model is built by model identification with measurement data to describe the dynamic shape recovery process of the 4D printed Shape Memory Polymer (SMP). Precise shape control is achieved and the effects of controller parameters on the precision of CL4DP are studied. Traditionally, SMP has a discrete number of selected steady states. With CL4DP, such steady states can be continuous and arbitrary.

With a higher requirement on machining accuracy in electrical discharging machining (EDM), there is a drastically increased number of multi-axis line segments for machining complex workpieces. High speed jumping is needed for a conducive discharge gap status. The increased number of line segments, however, makes the implementation of high-speed jumping difficult. To alleviate the online computational burden for velocity planning, this paper proposes a multi-axis velocity planning based on a coder-player architecture. Under this architecture, a multi-axis feeding trajectory is represented by a series of double non-uniform rational B-spline (NURBS) curves, which are then interpolated by the unit arc length increment interpolation (UALII) method. Based on the interpolation points with a resolution of a basic length unit (BLU), a velocity limit curve is generated by considering simultaneously the constraints on chord error, acceleration and jerk. The velocity planning is implemented under a coder-player architecture. As a trajectory represented by a double-NURBS curve has C2 continuity, velocity losses can be avoided at junctions between line segments, thus smoothing of line segments at junctions is unnecessary in jumping. Machining tests showed that by the use of the velocity planning based on double-NURBS curves, the machining time for one channel of a shrouded blisk can be decreased by 29.38%.

With the extensive application of 3D printing (3DP) in smartmanufacturing, 4D printing (4DP), which enhances 3D printed objects with shapemorphing ability by using smart materials, has shown significant industrial potentialand attracted tremendous attention. One key concern of 4DP is how to effectivelyand quickly meet different production and application requirements considering thecomplexity of materials and diversity of stimulus methods. In order to provide ageneral research platform for 4DP researchers, a flexible 4DP service platform isproposed. Components and modules for building 4DP and test systems are modeledand virtualized to form the different resources. These resources are then integratedvirtually or physically to provide some basic functions such as a 3D displacementstage or a visual monitoring system. According to different 4DP requirements, thesefunctions are then encapsulated into services to serve different research. Theplatform enables a variety of 4DP applications in smart manufacturing environmentssuch as 4D printed magnetic medical robots, test platform for studying the 4DPresponse, etc. A case study on designing a ferromagnetic 4DP platform based on theservice platform is performed to prove the feasibility of the method.

Converting tool path geometries of a computer-aided design (CAD) file to computerized numerical control (CNC) commands is the core functionality of a computer aided manufacturing (CAM) software for wire electrical discharge machining (WEDM). The number of curve objects (such as lines and arcs) for a complex tool path can reach more than 20000. However, curve objects in a CAD file are sorted by drawing sequence instead of processing sequence, and slow sorting will cause an increase in elapsed time (more than 120 seconds accordingly) and a hanging consequence to the running of the embedded CNC system. Finding adjacencies of curve objects is a key step in curve sequence sorting. Conventional algorithms targeted at adjacencies traverse all vertexes of CAD files, whose complexity is O(n2). According to the characteristics of a WEDM processing path, Ana-Hash Table Sorting algorithm is proposed to reduce the elapsed sorting time. This algorithm classifies vertexes into corresponding buckets in advance according to the coordinates, and adjacencies are judged among specific buckets, so that most useless traversals can be avoided. The complexity of the algorithm is O(nlogn). Noticing that the sorting acceleration ratio varies widely in table parameters, the optimization of table generating rules is given. The proposed algorithm works efficiently when the number of buckets per row is 1% to 1.5% of the number of curve objects. And the elapsed sorting time can be dramatically reduced by 99% as compared to the conventional traversal algorithm.

By the developing and the application of cloud manufacturing, many new business modes based on the IoT, CPS, cloud computing are springing up. It is worthwhile to model a new business mode simulate it to verify the feasibility. In this paper, with a demand of smart devices from a fresh-mix drink company, service agents are used for modeling and a basic comparison model is proposed for the negotiation between service providers to vote for the service provider that best suits the consumer's demand. A simulation platform that originates from cloud manufacturing is used to simulate the drink ordering, provider negotiation and provider recommendation. Simulation results show that consumers can obtain expected drinks after the negotiations and votes made by providers. Moreover, the system can recommend drinks with different freshness depending on the expiration parameter set by the company.

Soft robots, made of soft materials such as di-electric elastomer or shape memory polymers, have receivedtremendous attentions due to its dexterousness, flexibility andsafety compared with rigid robots. However, wider applicationof soft robots is limited due to their complex fabrication processand poor controllability. Here, we introduce a closed loopcontrolled soft actuator that is fully 3D printed with flexiblematerial. The structure of the soft actuator is optimized withFinite Element Method (FEM) to acquire shortest fabricationtime and highest deformation for same stimulus input. A desk-top Fused Deposition Modeling (FDM) 3D printer is used forlow-cost fabrication of such actuators. A webcamera is used forthe image feedback which offers the real time shape monitoringof the soft actuator. An output feedback Proportional IntegralDerivative (PID) controller with lowpass filter is developed withpole placement design method based on a data-driven modelof the 3D printed soft actuator. The controller is implementedto regulate the input air pressure to ensure a fast-response, precise and robust shape changing for any work environments.

CNC (computer numerical control) systems play an ultimately important role for controlling EDM (electrical discharge machining) machine tools and their machining processes. Till now, existing CNC systems do not offer sufficient openness that supports researchers and engineers to expend its capabilities and functionalities in response to the increasing demands of smart manufacturing; on the other hand, transforming an EDM machine made by small and medium-sized machine tool builders, into a smart manufacturing system has never been an easy job. To address the issues and overcome the difficulties which block the way towards smart manufacturing, this paper proposes an open architecture CNC platform for EDM machine tools. This platform utilizes the state-of-the-art technologies in implementation of the hardware and software without compromising with the constraints of obsolete techniques. For demonstrating the unique capabilities, the generalized unit arc length increment (GUALI) interpolation method and the Digitizer/Player system architecture are adopted. To exhibit the feasibilities of the newly developed platform, three kinds of EDM machine tools are applied associated with advanced functionalities such as machining process adaptive control, applications of machine learning, 6-axis EDM of shrouded turbine blisks etc. In addition, a small-scale smart manufacturing unit for drilling film cooling holes of turbine blades is built up into real production by applying the new CNC system and related software applications. From the practitioner's viewpoint, openness and standardization are the keys that enable the people from academia and industry bringing in their domain knowledge to enrich the smart manufacturing ecosystem.

The thermoplastic polyurethane (TPU) material is an elastomer that canbe used for inflatable products. Fused deposition modelling (FDM) is a widelyused additive manufacturing process for TPU material due to the capability ofgenerating complex structures with low cost. However, TPU is soft and thusdifficult to be extruded as continuously and uniformly as hard materials such aspolylactide by FDM. Inappropriate extruder structure and speed settings can leadto filament buckling problem, resulting in poor material filling quality, longprinting time and low printing success rate. This paper aims at improving the FDMprinting efficiency of TPU inflatable products by adding lateral support to thefilament and finding out the appropriate speed ranges for different wall featuresand thicknesses. Firstly, a filament guide sheet is designed as being inserted intothe gap between the drive gears and the bottom frame of the gear chamber in orderto prevent the soft TPU filament from buckling. Secondly, inflatable product wallfeatures are classified into floors, roofs and sidewalls and experiment for findingthe relationship between printing speed and airtightness is carried out. In order toverify the proposed solution, wall features are printed and the material fillingsobtained under different printing speeds are compared by measuring theairtightness of the wall features. Results show that the proposed filament guidesheet mitigates filament buckling, and the speed range that meets the airtightnessrequirement can be found for various wall features and thicknesses. In summary,the sealing of the filament feeding channel between the drive gears and the nozzle,as well as the speed optimisation according to product features, are essential forthe efficient printing of TPU inflatable products.