PolyGlove: a body-based haptic interface
Part 1: project profile
Project name
PolyGlove: a body-based haptic interface
Project short description
PolyGlove is a new kind of haptic interfaces, that try to overcome the borders of the present datagloves exploiting the EAP technology. This glove belongs to the body-based haptic interfaces, interfaces that use the user's body as force reaction base. EAPs, or better electroactive polymers, are materials that exhibit physical behaviors in response to an electrical stimulation, or vice versa, return an electrical signal in response to a physical stimulus. These materials can work as sensors and actuators, and thanks to their skills they can represent a solution at the previously mentioned limits. In this paper is presented the synthesis of a work where the EAP technology is applied into the haptic interfaces field.
People involved
Project head(s)
Prof.ssa Giuseppina Gini - User:GiuseppinaGini
Other Politecnico di Milano people
Ing. Paolo Belluco - User:PaoloBelluco
Students currently working on the project
Gabriele Valentini - User:GabrieleValentini
Part 2: project description
Introduction
This thesis project can be collocated into the robotics field, that science that proposes itself to integrate, in intelligent manner, perception and action. Into this field, the specific subarea of research is the haptic interfaces area, where are studied solutions to address the need of interacting with remote and virtual words, and in special way the project belongs to the body- based haptic interfaces. Thats devices that use the user's body as the force reaction base and have the inherent characteristics of being mobile, and therefore overcame the location dependency limitation. The purpose of this thesis is design and build a dataglove able to overcame the borders of the actual haptic interfaces. These borders are linked to the technologies used for develop the interfaces, that make difficult the integra- tion of force and tactile feedbacks into the same device. The main goal of the project is realize a device complete of force and tactile feedbacks, that at the same time guarantees wearability, comfort and movement freedom. The keyword of the solutions design is EAP, or better, the electroactive polymers technology. These new materials are plastics and gels thats show mechanical behaviors due to an electrical stimulation, and vice versa. They can be used as new kind of sensors and actuators, making possible overcame the limits of traditional technologies.
Brief description of the work
PolyGlove is the fruit of this research, a dataglove that base all his systems on the EAPs technology. The work did can be subdivided in several different 1.3. Structure of the thesis 2 phase. The first step was a global design of the whole glove, or better, was find a realizable solution for all the systems needed without project it in the de- tails. For the position/motion feed is proposed the use of two polymers: a piezoelectric PVDF film, and a piezoresistive CE silicon. This polymers can works as sensors, producing an electrical field (PVDF) or presenting changes in resistence (CE). For the force feedback, is proposed a double solution of active and passive feedback. Pistons full of ERF polymers and linked to the hand's fingers, can be driven in voltage in order to change the viscosity of the uid and to favor or obstruct the movement passively. The active side of force feedback is constituted by artificial tendons made of elastomer polymers. Concluding, the tactile feedback use PVDF as actuators in order to produce vibrations on the hand's palm. Special ciliary fingertips based on IPMC can return a texture perception. In a second phase, the design and develop process of the glove is started from the posistion/motion feed system. A first prototype is constituted by a glove equipped with four PVDF sensors, a control system able to sampling the sensors and to send the results to the remote side via wireless connection. This control system is made by different modules, in order to simplify the prototyping phase and to allow the reuse of the single modules. The brain of the system is a PIC18F2420 microcontroller equipped of an external 12 bit ADC, the wireless communications are guarantee by an XBee module, a switching voltage regulator manage the batteries and a customized sensor board process the sensors signals. When the prototype was ready, a phase of test on the glove is started. This tests were subdivided into hardware settings test, and sensor's characteriza- tion tests. The hardware tests make possible to find the proper configuration of several parameters like frequency of sampling, filtering and so on. While the practical tests have given results about sensor's behavior, showing good skills and evidencing some weak points of the system. At the end of this work, the critical conclusions have been made showing the good characteristics of piezoelectric films as motion sensors, and evidencing the problems of the system. In order to complete the work, the basic guide- lines of the future works are presented. This guidelines explains how to fix the problems and show the way of future developing works.
Future works
HW architecture
The hardware developed during this work present advantages from one side and disadvantages from the other side. The mainly advantages are the mod- ularity and the fast prototyping due to the technologies chosen. For instance, the control board is so general purpose that can be used in the future tests of the others feedback without modify it. The same reasoning can be did for the source and communication boards. The main disadvantage is the size of the circuit. Only to test a part of the position/motion feed, it been covered all the arm by electrical circuits. Thinking to the future, when all the feedbacks will be ready, the need of space for the glove will be great. An other point that need to enhancements is the main architecture of the modules. In order to minimize the electrical circuits, di�erent technologies must be considered. The Pin Through Hole (PTH) technology is very useful for a fast prototyping, but need to much space and limit the complexity of the boards. The solution is given by the Surface Mount Technology (SMT) that is a method to realize electronic circuits in witch the component are mounted directly onto the surface of the Printed Circuit Boards (PCBs). By the way, SMT give a lot of advantages like: reduce the space needed and the size of the components, the boards need to less energy for work, components can be placed on both sides of the boards, reduce the costs and so on. An example of the capacity of SMT is presented in the upper �gure. The switching regulator and all his components can be "compressed", thanks to SMT, into a box smaller than a stamp. The gain in space is evident, this regulator is 7-8 times more small than our regulator. Thanks to this new feature, is possible to realize small independent modules for each feed/feedbacks. Every modules can be supplied by a microcontroller dedicated only to that purpose. Following this way, is possible to design the modules in a di�erent manner. For instance, the position/motion feed don't need to a microcontroller so big like PIC18F2420, but it can be replaced by a smaller one. This choice reduce the computational load for every microcontroller exploit- ing the parallel calculate. On the head of the varied modules can be placed the Xbee. In fact, the Xbee is not only a wireless system, but it is supplied by several analog and digital lines. Designing a proper circuit, the Xbee can work like router for every modules, analyzing the packets and addressing their to the right modules. Concluding, the hardware need to some improvements. SMT technology simplify the electrical circuits, and make possible realize small modules for each feedbacks. Using the Xbee like router for the modules, the future hard- ware will be composed only by a module for position/motion feed, one for force feedback, one for tactile feedback, the SMT regulator and the Xbee. If the battery pack will be decentralized onto the high part of the arm, the whole structure can be placed in a small bracelet.
SW Architecture
At this point of the project, PolyGlove is not supplied by a proper software architecture, but it work only with simple MATLAB commands. MATLAB is surely the best choice for the prototyping phase of the project, but when the feedbacks will be ready, PolyGlove will need to something much articu- lated. The �rst need after have complete the feedbacks will be a solid commu- nication protocol. When the Xbee will work as router for the modules, the communication protocol will be fundamental for the whole device. The protocol must be light, with a mechanism of errors control and stable. Poly- Glove, when ultimate, will need and will produce a great amount of data, and this data will stream through the XBee. The second need of the glove will be a low level API for communicate with the device. Basing the architecture onto open source API like Chai3d, the future works of this side will be develop a low level architecture able to link the hardware with the hight level software, in order to provide an easy way to write demos for the glove.
Position/Motion feed
How viewed from the tests, the part of position/motion feed developed need to improve same points in order to solve problems. From the other side, there are a lot of new works to do for �nish the developing process of the feed. In the next rows will be presented the guidelines for �x the problems coming from the tests and for �nish the undeveloped points of the feed. Sensor's shape and size. As viewed in section 6:2, Polyglove su�er of some problems due to the improper shape and size of the PVDFs, prob- lems such the reciprocal disturb of the sensors. The sensors used have a rectangular shape of about 2:5 per 1 cm, and these shape is the cause of the problems. Piezoelectric �lm is available in a va- riety of di�erent �lm sizes and thicknesses. These can be fabricated into simple transducers, or for use as full size sheet for other application. This virgin sheets are available without electrodes attachments and wires. Start- ing from these sheets, instead from prefabricated sensors, make possible to realize customized sensor's shapes. If rightly positioned, the sensors can be drastically reduced in size. Reducing the sizes and customizing the shapes can solve the reciprocal dis- turb problem. By the way, this choice improve the glove performance, reduce the cost about sensors and make possible realize sensors for all the �ngers. PVDF characterization. Another point that need to be complete is the PVDFs characterization as sensors. The future works should be con- tinue the study on the PVDF skills and �nd a proper mathematical model for transform the voltage level into position/motion informations. In order to complete the characterization, a secondary position system is needed. Several solutions are possible: mechanical structure with poten- tiometers, piezoresistive bend sensors, hall e�ect sensor and many others. Due to the only momently need of this system, the technology chosen is not very important, but design a structure that will not in uence the PVDF behaviors is the real goal of this future work. When the supporting system will be ready, the process of characterization will continue and the next step will be �nd a mathematical model for the sensor. A way for do this is constituted by the neural networks. This math- ematical model can help the characterization process. By the way, with the supporting position system the developing process of the networks will be automatized in his major part. CE polymers. Another future work is constituted by the piezoresis- tive CE. How just said in the previous chapters, this polymer make possible realize a variable resistive circuit on the upper side of the hand. The future works along this way will be realize a prototype for the glove and start a mathematical analysis of his work. The weakness of this polymer is the slowly change of resistence due to a movement. This skill make it perfect as statically position recognizer but make di�cult use it during the dynamic phases. Now the situation is: from a side the fast and dynamic PVDF �lms and from the other the slow and static CE silicons. Integrating the data from PVDFs and CEs can make possible realize the �nal position/motion feed system. By the way, the data coming through the CEs can be used directly into the PVDF's neural network realizing a unique mathematical model, or can be created two di�erent preprocessing models for the systems and a third mechanism that take and merge data for the �nal use.
