Multi-university design projects
We describe a multi-university design project in which teams of students across different campuses collaborate on a design and manufacturing project. We show how such projects sensitize students to issues in concurrent engineering and train them in interpersonal skills, communications, and system integration. We believe that this approach allows us to simulate real-world conditions by imposing realistic boundary conditions on the student teams.
An important aspect of real-world design and manufacturing is the requirement to work in teams.’ An important aspect of this requirement is the ability to define where one team member’s responsibility starts and where the other members’ responsibility ends. This process of breaking down a design and fabrication project into tasks and defining responsibilities forces the team members to think in terms of subsystems and to address the integration of subsystems. Generally, the only attempt to incorporate real-world considerations into the curriculum comes in the capstone design project in the senior year. Here team design projects do emphasize the importance of working in groups and developing interpersonal skills, but this is often contrived. Frequently, one or more team members assume total responsibility for the project, and this defeats the objective of team projects.
In an attempt to bring real-world challenges to design projects, we have, over a period of four years, conducted multi-university capstone design projects. The team is assembled from students from more than one university campus. The campuses are geographically distributed, thus simulating the real-world situation in which product development team members may live in different cities or even countries. This makes the team members break down a project into subsystems that are loosely coupled and can be designed more or less independently by a subgroup at a single campus, and address the integration of subsystems into the final product. The expenses involved in traveling and time conflicts make frequent meetings difficult. This forces the students to alternative methods of communications (such as electronic mail, Internet, and video conferencing). Ideally, the subgroups in different campuses have different expertise, and the collective expertise of all team members is required for product design and development. Finally, the students learn concurrent engineering, the practice of integrating marketing, manufacturing, and design issues.
The schools participating in our concurrent design projects included The University of Pennsylvania (Penn), The Ohio State University (OSU), The Cooper Union (CU), New Jersey Institute of Technology (NJIT), and Drexel University (DU). The teams consist of mechanical engineering and industrial engineering students. Thus, while this paper addresses some fairly general issues in engineering education, the specific projects have significant mechanical design and manufacturing content.
The projects described here were conducted over a period of four years. In the first year, this experiment was carried out between two universities, Penn and OSU, with five students across the two campuses. As table 1 shows, the projects grew in size and overall scope. In the fourth year, we had five participating universities with as many as twenty students.
II. THE DESIGN PROJECTS
A. Keychain Charm
In this one-semester project, seniors from Penn and OSU designed and fabricated a 5 cm by 7.5 cm key chain charm with the letters PENN and OSU on it. Penn students, in consultation with OSU students, designed an aluminum mold for the charm. The CAD model for the mold was used to generate CNC machine code, which was first used for prototyping a scaled down machineable wax model at Penn, and then shipped over the Internet to OSU. OSU manufactured the mold and used a desktop injectionmolding machine to make several copies of the charm.
B. Writing Linkage
In this one-semester project, students from Penn, OSU and CU designed and fabricated a writing linkage. The linkage is an assembly of rigid links, cams and springs that can be driven by a single crank to write a pre-determined script on a 5 cm by 7.5 cm card. The design consists of two concentric cams and a five-bar linkage. The cams, driven by an input crank, drive the two inputs to the five bar linkage (figure 1). More details are provided in Table 1 and Figure 1.
The brainstorming and preliminary design stage mainly involved the Penn and the OSU teams. Penn students developed the first design with comments and suggestions from the other schools, particularly concerning the manufacturing issues. The resulting design was shared with CU and OSU students. CU students performed a simulation and analyzed the design for errors. OSU students investigated the manufacture of the product in greater detail. Penn and OSU students worked together to develop the second design that overcame the flaws pointed out by the analyses of OSU and CU students. This process was repeated two more times before the first prototype was manufactured.
C. Feeding Device for Quadriplegics
This was our first two-semester project, and it involved students from Penn, OSU, CU, and NJIT. It was also our first attempt to include conceptual design and product development in the scope of the project. From the very outset, students who were familiar with market analysis and product development contributed to the project.
There are many individuals with severe physical disabilities who cannot feed themselves. Although there are commercially available feeders, they tend to be expensive or difficult to adapt to individual needs. A recent survey of the U.S. population indicates that approximately half a million people could benefit from well-designed feeding aids. Because of its humanitarian aspects, this project had a great deal of appeal to the students.
Based on a survey of consumers and practitioners, the students arrived at the following design considerations for the feeding device: (a) low cost, (b) ease of use; (c) safety; (d) aesthetics; (e) simplicity; (f) portability;and (g) low weight. Further, it was decided that the device should be passive and completely mechanical (i.e., without any electrical components). Such a device is likely to be more acceptable and easier to use than electronic ones.
Different candidate designs were pursued independently by the four schools for one semester. At the end of the first semester, the students agreed on a design that combined the best features of each design. The second semester was spend in refining the final design, analyzing it using computer simulations and animations, and building a prototype.
In figure 2, the final prototype is shown being demonstrated by one of the participating students who was a quadriplegic. The device consists of a feeding utensil that is supported by a counter-balanced arm. The user grips one end of the feeding utensil to which a mouth piece is attached and is able to scoop up food from the plate. The plate consists of compartments. The user can rotate the plate about a vertical axis with the help of the utensil. The frame with circular arcs assists the user in manipulation. A key aspect of this design is the utensil holder that consists of a spherical joint with a clutch. The user can deactivate the clutch to scoop up food or to manipulate the plate. However, once the food is scooped up, bringing the spoon (utensil) to a horizontal plane activates the dutch. The clutch keeps the spoon (utensil) horizontal, allowing the user to release the spoon and rotate it so that the spoon faces the user. During this process, the planar arm assembly, which consists of links, pin joints and springs, keeps the spoon assembly counterbalanced against gravity.
In the overall project, Penn students designed and manufactured the arm assembly. OSU students designed the utensil holder assembly that consists of the supporting ball joint, the dutch, and the frame with circular arcs. CU students designed and fabricated the table and the rotating tray, and NJIT students designed and fabricated the bowl. These subsystems are shown in figure 3.
In three months, all of the teams were able to design and produce prototypes of their subsystems. They met to inspect the prototypes and to discuss the assembly issues. This meeting revealed several design flaws, and the teams had about a month to fix the defects. In the next six weeks, each team revised their designs and produced four prototypes of each subsystem. Four working assemblies were produced and demonstrated. The device was found to be easy to use and manipulate, and not much physical effort is needed for its operation.
D.An Assistive Robot Arm for Wheelchairs Users
The goal of the project was to design and manufacture a wheelchair robot arm for people with disabilities. In Phase I (approximately three months), each university team developed a conceptual design independently. The teams met and discussed each design at a January meeting at Cooper Union and developed the conceptual design for the final product based on the individual team designs. This design formed the basis of the final prototype, which was developed and prototyped in Phase II.
A market survey indicated that there are as many as 3 million Americans who are wheelchair users and who might benefit from a personal robot assistant. An estimated 200,000 are children who need personal manipulation aids. A’robot that can be used by a wheelchair user will directly serve this population.’ It will be useful not only in activities of daily living but also in learning, playing and in hobbies and crafts. There are a number of rehabilitation robots currently available or in development. However, these are either very expensive (over $40,000) or difficult to use. Therefore, the students could make a contribution to society if they could design a relatively inexpensive, easy-to-use robotic arm.
In a September meeting, team members from five schools developed the following design specifications. The arm must be usable by children. It must have a reach of at least 1 meter and a payload of at least 10 Newtons. The grip force must be less than 10 Newtons. The gripper aperture (also the width of the grasped object) must be at least 7.5 cm. The speed of the end effector must be limited to a maximum of 2 meters/second and the minimum speed must be 0.1 meters/second. Finally, the user must be able to pick objects off the ground, and the arm should be controllable from an IBM-compatible PC. The net cost of all the parts excluding the computer but including machining costs must not exceed $4000.
In the first semester, each school pursued independently a conceptual and preliminary design for the robot arm. Each design was analyzed to the extent necessary to prove its feasibility. The final design was chosen in a January meeting. Once again, the best design was one that combined the best features of all the individual designs. The second semester was spend in refining the final design, analyzing it using computer simulations and animations, and building a prototype. The final design consisted of a five degree-of-freedom robot arm with a torso/waist, shoulder, elbow, and wrist. The wrist has two degrees of freedom, pitch, and roll. The gripper is a parallel jaw gripper with an on-off control. All the joints are driven by DC electric motors through transmission elements designed so that the motors can be placed on the base, thus making the inertia of the moving parts as low as possible. The transmission includes safety clutches to ensure large forces/torques cannot be applied by the arm. The control is through a standard, video-game joystick. The easy-to-use joystick controls the manipulator end effector in Cartesian mode. The joystick commands are read through the game port of a personal computer. Appropriate software decodes these commands and translates them to desired Cartesian velocity commands. Another level of software is used to translate Cartesian velocities into joint velocities, to compare the desired joint velocity commands to the actual joint velocity, to calculate the discrepancy, to calculate the desired motor input signals, and to write a proportional command to a motor chip. A control circuit interprets this command and generates the desired voltage and currents to move the motor.
In this project, DU students were responsible for the arm housing and skeleton. OSU students designed the transmission, and Penn students designed the real-time control system and the actuation package. CU students designed the user interface, and NJIT students designed the gripper and conducted market surveys. A picture of the final prototype is shown in figure 4.
III. THE PROCESS
A The Different Stages ofthe Design Project
The phases of the projects are identified in table 2. There is a meeting associated with each phase as shown in table 3. The first and most important part is the initial brainstorming involving all of the students. This leads to the problem definition and formulation of the statement of need. At the end of this stage, the goal of the project has been defined and many critical (but not all) specifications have been agreed on. This takes anywhere from a week to two weeks, but most of it is accomplished at the first meeting of all the participants. The technical and business aspects of the project are broken down into as many areas as the number of participating schools. For example, in the case of the assistive robot arm, the five technical areas were real-time control and sensing, user interfaces, simulation, transmission, and market analysis.
The second part of the project takes the form of a design competition, in Which school works independently to come up with the best possible design. While the teams are encouraged to share the technical information, market analysis research, and data sheets of products, there is a healthy competition among the teams as far as the design is concerned. At the end of this phase (roughly four months), all schools have a candidate design. They write a report describing their design with analysis to demonstrate the feasibility and the main advantages.
The third part of the project consists of evaluating the candidate designs and determining the best possible design. This best design generally ends up including the best features of all candidate designs. While the students are generally very good at identifying weaknesses of their own designs and acknowledging the strengths of other designs, it helps to have the unbiased advising faculty members lead this process. This phase consists mainly of project planning. The task of developing the details of the final design and manufacturing it is broken down into subtasks and each school is assigned one or more subtasks. Often these subtasks correspond to designing and making subsystems (or subassemblies) of the product. This phase also includes the preparation of a detailed schedule for completion of subtasks (subsystems) and deciding on deadlines.
The fourth part of the project, detail design, consists of preparing detailed designs for each subsystem. Each school focuses on the subsystem assigned to it. But since a subsystem is generally coupled to other subsystems, the teams have to communicate and collaborate extensively. This part of the project is characterized by design reviews in which each team presents their progress. This phase generally takes between 10-12 weeks. At the end of the project is the final design review by which time all teams have functional prototypes or at least mock ups of their subsystems. At this review, the teams try to put together a functional system from the subsystems. However, this is only possible if the designs are perfect. Because a different team designs each subsystem, the teams discover many design flaws, some stemming from manufacturing considerations.
The last phase is the redesign, manufacturing, and assembly phase. In this phase, the final iteration of the design is completed. At the end of this, each team will have manufactured prototypes of a subsystem and will be ready for assembly. The assembly is done at a final meeting that is held at the end of the project. The students are able to assemble the prototype and demonstrate it. They also make a joint presentation to the faculty and external reviewers.
There are four meetings at which all the team members try to be present. The kick-off meeting is the first meeting at which the teams collectively decide on the overall goal of the project, the problem statement, the scope of the project. All team members are present at this first meeting. (In subsequent meetings, each school (team) may choose to have their key representatives participate in order to reduce costs.) The kick-off meeting also marks the start of the design competition. After the design competition, the schools meet to arrive at the specifications of the final design. They also break the project down into subsystems and subtasks and assign responsibilities. The faculty advisors assist in this process. The assignment of responsibilities is based on the collective interests of each team, their expertise, the facilities available at each school, and the expertise of the faculty advisor. The faculty advisors, with their knowledge of the team members and the educational training received by the students at each school, play an important role in this process. Our experiences with the different phases and a schedule of meetings are summarized in tables 2-3.
IV. ADMINISTRATIVE AND ORGANIZATIONAL ISSUES
A. Communication Among the Teams
One of the main goals of the project was to develop the skills, which would allow members at different institutions to successfully work together to concurrently design and manufacture a device. To properly convey ideas to effectively design, analyze, re-design, and later manufacture of the device, it was essential to explore and employ a number of different communication channels.
The most important means of communication by far were the meetings in which all of the schools traveled to one location to discuss ideas. These meetings were critical to the project because feedback about the designs was quickly exchanged from all schools. This is especially true of the third meeting when the pre-prototype was presented. The parts developed by each of the schools were assembled to form a working model, which permitted us to ascertain the highlights as well as the shortcomings of the model. Although a meeting of all four schools at a single location was very productive, it was by far the most expensive means of communication and could not be done often. The next best alternative was also the most innovative means of sharing ideas: video conferencing. Each of the participating schools purchased a video conferencing system.
Once the difficulties of arranging common meeting times and the initial hardware and software problems were overcome, video conferencing worked very well. The system permitted the visualization of ideas, which would otherwise be difficult to convey verbally. For example, when the Penn, OSU, and DU teams were discussing the arm and transmission, they quickly sketched pictures and with gestures were able to explain a difficult idea quickly and easily. When a question or a confusing issue arose, it was resolved right away. After the systems were operation, the students scheduled at least one videoconference session each week.
Data transfer was accomplished mainly via the Internet and facsimiles. E-mail was heavily utilized for transfer of data files and mass communications. Schematics, MS Word documents, and MS PowerPoint presentation slides were encoded and attached to a letter. The e-mail address of every member of each team was distributed at the beginning of the project. In addition, a common address, or alias, was established so that mail could be sent to all members of every team at once. This was a very efficient way to communicate issues to the group as a whole.
Each school hosted easily accessible project homepages on the World Wide Web (WWW) which had information such as design information, updates, weekly reports, e-mail addresses and phone numbers of all the people involved. The Web was useful to show pictures quickly without having to e-mail them to the other schools, which would then have to translate the file to be viewed on the computer. However, keeping the homepage updated was time consuming, so facsimiles were used mainly as a quick way to transfer pictures in between video-conferencing times and also as a supplement, right before video conferencing so that the design could be discussed in detail.
For brief communications, the telephone was used. To enable multiple members of a team to communicate, a speakerphone was purchased.
The main difficulty in the logistics of organizing such a multicampus design project stems from the difference in university calendars. In our case, two schools (OSU, DU) are on a quarter system while the others are on a semester system. The schools start at different times and the spring breaks do not coincide. Some schools (e.g., OSU) finish their spring term in mid-June, while others (e.g., Penn) start their summer break in May. Further, Penn’s capstone design project course runs from the junior spring semester through the senior fall semester while all other students did their capstone design project in their senior year. We had to find a way to work around these differences and the need to grade the students according to their school’s calendar (particularly for graduating seniors). Most schools used paid summer jobs as a way to entice students to come early or stay after the end of the school year. Independent study courses were used as a vehicle to customize the students’ course plans to allow them to participate in the projects. Recruiting students for the project had to be done before the school year, often during the previous school year.
The second difficulty that arises in multi-university projects is the lack of proximity among campuses. Travel between campuses, we found, was absolutely essential. As pointed out earlier, we believe it is necessary for all the students to meet each other at least once. Even though not all students attended the other meetings, the travel costs are fairly significant. In our case, often the travel costs were minimal for Penn, NJIT, CU and DU, but it was more difficult for the OSU team who had to travel to the Philadelphia or New York area for the meetings. It cost us $5000-$10,000 per year to pay for travel.
It is important to set up the basic infrastructure for communications well before the project is initiated. This consists of
Setting up a distribution list for electronic mail; Establishing websites for each school; and
Setting up and testing a video conferencing system at each school.
Of these, the last is the most difficult. Often this requires an investment. on part of the participating university in terms of the desktop equipment and/or communication lines (e.g., ISDN phone lines).
Our evaluation of the projects was mainly through exit interviews using a standard questionnaire. After the project, we asked the students what they had learned from the projects and to compare their experience with those of their colleagues.
There were three dear benefits to the multi-university design projects.
1. Sharing of facilities: Students were able to use laboratory and experimental facilities at other institutions. For example, NJIT has an excellent manufacturing facility and all students were able to use NJIT’s facilities during the collaboration. Similarly, in the last project, all the machining and assembly was done at DU’s easy-to-use machining center.
2. Lesson in concurrent engineering: The design projects were useful in training students to work cooperatively in a concurrent engineering framework. More importantly, the students were sensitized to the concept of remote, real-time manufacturing and to the potential of computer integrated information technology. The project also gave them a sense of the real world in which vendors, designers and manufacturers from different parts of the country (or the world) have to collaborate to realize a product.
3. Lesson in system integration: A key aspect of this project was the geographical distribution of the different teams. The teams could not simply meet on a whim to resolve questions or arguments. That meant the students had to use the available meeting time effectively and present their ideas in a clear, concise fashion. Students also learned that describing the characteristics of their subsystems and obtaining the specification of other subsystems is extremely important for the downstream integration. It is very difficult to teach the intellectual aspects of system integration. However, it is our belief after the exit interviews, that this is a great way to teach system integration.
Other intangibles have become apparent over the last several years. First, students learn to respect each others’ time and to be disciplined. In an inter-campus collaboration, it is unusual for two collaborating teams to have exams at the same time or to have spring breaks at the same time. Thus, time conflicts were a rule rather the exception. The students learned to compromise, make time for meetings, and respect deadlines.
The second intangible benefit has to do with the interaction among the students. There was always a mix of students (different ethnic, gender, and cultural backgrounds) working together. Also the participating schools are very different. For example, OSU is a large state institution while Penn is a small, private school. CU is a small school with an emphasis on undergraduate education but without a significant research program. NJIT has a large part-time student population, a feature not shared by any of the other schools. DU is a strong engineering school in contrast to Penn, which prides itself on a solid liberal arts education. Thus the students in these institutions come from different backgrounds and are schooled very differently. Therefore, they also learn a great deal from each other.
Finally, the competition among schools brings out the best in all the teams. In the design competition phase, this was very obvious. Students at each school worked hard at their design and also their presentation. But even in the collaborative phase, no school’s team wanted their part to be the weakest link in the chain.
During the exit interviews, some of the disadvantages of the multi-university project concept also surfaced. First, there is a tremendous overhead associated with collaboration. Often there is a feeling that one is better off doing the whole project on her own. Some students felt they wasted a lot of time trying to guide the project in the right direction and did not feel that they got the cooperation they deserved. But these concerns are not unique to multi-campus projects; they frequently arise in any team project. Perhaps the frustration is exacerbated by the fact that students from different campuses often have different priorities. Again, this is because of the different backgrounds (educational and cultural). The real disadvantage of the multi-campus project concept is perhaps the time and energy spent in communication. The desktop video conferencing technology is still not at a level where it can be used as a telephone. The process is particularly frustrating because of the poor frame rate, the inadequate resolution, and the low bandwidth.
Another aspect that the students found frustrating had to do with technology. Each school advocates and trains students on a different computer system (for example, Macintosh, Wintel machines, and UNIX workstations). Often too much time was spent discussing how to get presentation files into the correct format or how to include figures from different sources (for example, solid models, scanned drawings, CAD blueprints, and photographs) into reports.
While our multi-university projects have been very successful, and we have steadily expanded the scope of the project and increased the level of participation, our most recent project involved five schools and twenty students. We believe that the logistics would make it difficult to include more than five schools. However, it is certainly possible to increase the number of students participating in such projects. One avenue is to include two groups of students, each consisting of teams from different universities, across five universities. This will allow two different groups to work, if necessary, on the same problem, and will double the level of participation. It is possible to have many groups. If they all work on the same basic problem, the overhead in involving more students does not become prohibitively high. Another approach is to increase the number of students working in a team. This is likely to work if the project is sufficiently large. An example of a project with this scale is the Formula SAE car. But even in such projects, it is difficult to include more than 30-35 students in a single project.
The issue of funding for multi-university design projects is a very important one. The natural question to ask is if external funding is required to sustain such a project. The answer is definitely yes. However, the question of the amount of required funding has to be addressed more carefully. If a school has the required infrastructure and a set of sister institutions, in steady state, the level of required funding is quite modest. In our experience, the main expense is the travel expense and this is generally between $5000-$10000. This of course assumes that there is university finding to support the use of computer and telecommunication facilities, the machine shop and other equipment. However, there are several approaches to sustain this program without funding from federal organizations. The obvious approach is to involve industry as a partner and address problems that are relevant to their needs. Another solution is to sell products as a non-profit corporation to make enough money to sustain the project. However, if this approach can be scaled to accommodate the entire class, and if its impact is as significant as it promises to be, there is a very good chance that this funding may come from universities. We only have to look at the Mini-Baja and the Solar Car competitions for examples of institutional commitment.
It is always difficult to evaluate the effect of new educational initiatives. Our evaluations were based on interviews of participating and non-participating students. However, we did not attempt to define any quantitative measures of the impact of multi-university design projects. Further, it is important to note that our evaluation was done with a small pool of students. It would be ideal to evaluate a sample of students who completed the multi-university projects and another sample of students who instead completed the regular capstone design project and compare them. It is difficult to arrive at suitable metrics that can be used to measure the effectiveness of the multi-university projects. Even if this could be done correctly, our sample size of 20-25 students is much too small for doing any quantitative analysis. Clearly, when this project is scaled to reach a larger fraction of the graduating class in each school, and a larger number of schools, a more quantitative evaluation will be feasible.
Finally, one important issue that was never clearly resolved is the patentability and ownership of patents of novel ideas. During the projects, students came up with new ideas that are patentable. However, it is not clear who would be the owner of these patents. It is possible that all the participating universities can jointly own the patents. Presumably, in this scenario, all the universities would contribute equally to the expenses involved in filing the patent. This was simply not explored in any detail.
The support of the National Science Foundation through the Gateway Engineering Coalition (grant EEC-9109794) is gratefully acknowledged. We thank the students who worked on their project for their unflinching enthusiasm in participating in this experiment. Unfortunately they are too many to list here. We want to particularly thank the graduate students who helped in 1996-97 in the most ambitious project: Tom Sugar (University of Pennsylvania), Spiros Koulas (Drexel University), and Tanya Schneider (Ohio State University).
*Includes time during term breaks.
1. Mechanical Engineering Undergraduate Education for the Next Twenty Five Years, A Report on a Workshop for U.S. Mechanical Engineering Departments, MIT, Cambridge, MA, 1996, Oct. 7-8.
2. Wei, C.S., V. Kumar, and G. Kinzel, “An Educational Experiment in Teaching Mechanism Design and Manufacturing Using Multi-University Teams,” Proceedings of the 4th National, Applied Mechanisms and Robotics Conference, Cincinnati, 1995, Dec. 10-13.
3. Biswas, A., et al., “Basic Design Optimization of Mechanisms,” Proceedings of the 4th National Applied Mechanisms and Robotics Conference, Cincinnati, OH, 1995, Dec. 10-13.
4. Prior, S.P., “An Electric Wheelchair Mounted Robotic Am: A Survey of Potential Users,” Journal ofMedical Engineering and Technology, vol. 14, no.4,1990, pp.143-154.
VIJAY KUMAR Department of Mechanical Engineering University ofPennsylvania
GARY KINZEL Department of Mechanical Engineering The Ohio State University
STAN WEI Department of Mechanical Engineering The Cooper Union
GOLGEN BENGU Department of Industrial Engineering New Jersey Institute of Technology
Copyright American Society for Engineering Education Jul 2000
Provided by ProQuest Information and Learning Company. All rights Reserved