Robotics Laboratory Classes for Spatial Training of Novice ...

International Journal of Engineering Education Vol. 31, No. 5, pp. 1376–1388, 2015 Printed in Great Britain

0949-149X/91 $3.00+0.00 # 2015 TEMPUS Publications.

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students* IGOR M. VERNER** Faculty of Education in Science and Technology, Faculty of Industrial Engineering and Management, Technion—Israel Institute of Technology, Israel. E-mail: [email protected]

SERGEI GAMER Faculty of Education in Science and Technology, Faculty of Industrial Engineering and Management, Technion—Israel Institute of Technology. Israel. E-mail: [email protected]

This paper presents a study in which learning practices of novice engineering students when programming and operating robot manipulators focus on the development of spatial skills. To provide the practices, we customized the laboratory setup: unified workspaces of available robots, designed virtual robotic cells, and developed robot manipulation tasks with oriented blocks. We examined outcomes of the proposed practice for two categories of learners: 248 first-year Technion students participated in the robotics workshop as part of the introductory Industrial Engineering and Management (IEM) course; 20 tenth graders majoring in mechanical engineering at a vocational high school, who took an outreach course in our lab. With regard to the latter, evaluation focused on the development of spatial skills and indicated significant gain in the performance of spatial perception, mental rotation, and visualization tests. For most of the Technion students the workshop was the first experience in robotics that aroused their awareness of spatial skills required to operate and program robot systems in manufacturing processes. The results argue for the educational value of the proposed learning practice and its further exploration in different settings. Keywords: CDIO approach; industrial robotics laboratory; introductory robotics course; training spatial skills

1. Introduction Introductory engineering courses (IECs) combine study of basic concepts with practice using tasks that resemble real problems faced by engineers and technicians in their professional lives. Such practice exposes students to the nature and challenges of the future profession, and fosters their motivation to learn. Many engineering programs, including those in industrial engineering and management, employ IECs to help first-year students explore their choice of an academic major and improve retention rates [1]. IECs are also delivered through outreach to prospective students, with the aim of stimulating their interest and enrollment [2]. A large and increasing number of engineering programs focus their IECs on conceive—design— implement-operate (CDIO) experiences [3]. Through CDIO assignments and projects, students learn fundamental engineering principles and apply them to design, implement, and operate technological processes. This approach emphasizes the role of laboratories in fostering creativity, learning motivation, teamwork skills, and ‘‘understanding of the engineering discipline prior to choosing an area of study’’ (ibid, p. 106). It calls to reorganize laboratories, traditionally oriented to advanced

** Corresponding author.

1376

engineering courses, into modular workspaces for all levels of engineering education. The ways to increase the efficiency of learning practices in Robotics and Computer Integrated Manufacturing (RCIM) laboratories are widely discussed [4–5]. For the majority of students participating in IECs, the laboratories offer their first exposure to manufacturing processes and the experience of operating robots. RCIM laboratories are workspaces in which students can explore modern production technologies and design solutions for various manufacturing scenarios [4]. These scenarios follow the human-in-the-loop model, meaning that the production process is carried out through active interaction between the human and the robotic system [6]. In education, the student-in-the-loop system is designed in such a way that learning factors are maximized [7]. Notably, while the human-in-theloop model is commonly used for training technical staff and practitioners, there is a substantial difference between staff training and educating unprepared students. Whereas the former focuses on imparting technical skills needed to work efficiently with specific systems, students need to learn the principles of robot programming and operation, and to develop generic abilities and skills that are required in different workplaces [8]. Among the most important of these is the ability of spatial vision. * Accepted 2 June 2015.

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

Industrial robotics laboratories, which are of interest in our study, generally implement three types of learning scenarios [9]: setting up a robot system, programming different industrial robots, and performing advanced robot-handling tasks. These laboratories offer learning practice in hands-on, virtual, and remote environments [10]. In hands-on environments, students are present in the laboratory and directly control the installed robot systems. In virtual environments, they work with computer simulations of robot systems. In remote environments, the students are distant from the laboratory and control the robot systems through teleoperation. To perform robot system setup, programming, and operation assignments, the student needs immediate and detailed visual information from the robot workspace. In the hands-on environment the student is near the robot system, and so all needed information is acquired directly through observation. In the remote control system, visual feedback is transmitted from video cameras via a computer screen, and so is incomplete and delayed [11]. In the virtual environment, the student works with a graphic simulation of the robot system on the computer screen, under limitations of the given software [10]. The advantages and constraints of the hands-on, virtual, and remote learning practices have been discussed and compared in the literature [10, 12– 14]. Less attention has been paid to the analysis of difficulties that students face while performing tasks in robotic environments, and to the impact of this practice on the development of fundamental engineering skills, including spatial skills [11, 15]. The questions that need further investigation include:  How can different environments be combined to support learning practice in the industrial robotics laboratory for both advanced and novice learners, as recommended by the CDIO approach?  What learning practices for novices can best foster the development of spatial skills and awareness of their importance in industrial robotics? The current paper reports on the results of our study conducted in the RCIM Laboratory of the Faculty of Industrial Engineering and Management at the Technion—Israel Institute of Technology. Over four academic years (2011–2012 to 2014– 2015), we ran in the laboratory three robotics workshops (course units) for IEM first-year undergraduates, and separately, three short outreach robotics courses for 10th-grade students in an underprivileged vocational high school. Both sets of courses offered learning practice in programming and operation of robot manipulators, fostered students’

1377

spatial skills and awareness of the importance of these skills in industrial robotics. To maximize the learning effect, we combined robot operation tasks in virtual, real, and remote environments. The environments and tasks were designed to provide targeted training of different spatial skills. Below, we first review the existing literature on spatial learning in robotic environments (Section 2). We then describe the modifications we made to the Technion’s RCIM laboratory in order to make it suitable for learning practice by novice students (Section 3). In Sections 4 through 6, we describe the study framework and the two case studies, in which we engaged the students in performing puzzle exercises with robot manipulators in hands-on, virtual, and remote environments. We report that learning practice in the robotics lab influenced the improvement in performance of spatial tasks (case study 1) and awareness of the need of spatial skills in industrial robotics (case study 2). We conclude by discussing implications of the findings.

2. Literature review: Spatial learning in robotic environments Engineering practice depends on visual information, and strong spatial perception, reasoning, and visualization skills are critical to success in engineering careers [16–17]. This is true for practice in design and operation of automated manufacturing systems (AMS). Indeed, with the rapid progress of AMS, the functions of the human operator are becoming increasingly intellectual and based on abilities in which humans trump machines. Engineers responsible for the design, operation, and supervision of AMS must have aptitude in dynamic perception and dynamic and flexible reasoning, as well as a capacity for autonomous work and for rapid yet accurate decision making [18]. Strong spatial skills are crucial for all aspects of robot design and operation, whether hands-on or remote. Lathan and Tracey [19] showed that performance in teleoperating a robot through a maze using a single camera significantly correlated with performance in standard spatial reasoning tests. Menchaca-Brandan et al. [20] found spatial skills, particularly perspective taking and mental rotation, to be essential for operating robotic manipulation systems. Spatial ability is even important in programming. Jones and Burnett [21] found a positive correlation between programming skills (debugging and code navigation) and mental rotation ability. Spatial skills can be developed through experience and practice [16, 22], and studies in spatial cognition suggest that digital technology environments can facilitate effective training in these skills

1378

[23]. Further, many studies have shown that spatial training can be incorporated into existing programs in STEM education, and that the impact of such training can be long-lived [24–28]. For instance, Sorby [29] found that including activities directed toward the development of spatial skills in undergraduate engineering graphics and CAD courses significantly improved students’ course achievement and their spatial reasoning and visualization skills. Similar effects were observed by Hsi, Lynn and Bell [30], who included spatial learning tasks in their introductory engineering course. Interactive simulations are increasingly used to facilitate spatial learning in engineering and science laboratories, and their educational value has been widely discussed in the literature [31-32]. Studies suggest that the effectiveness of spatial learning depends on characteristics of the virtual environment and on learning conditions. Kozhevnikov et al. [33] found that results of spatial learning in 2D and 3D non-immersive environments were almost the same, while learning in the immersive environment produced the best outcomes. However, we can likely expect different results in the case of practice with virtual robot manipulators. At the current level of robotic simulations, students who operate the mechanical arm in an immersive environment can face difficulties in egocentric spatial processing [34], as this operation requires the student to ‘‘put’’ himself/herself in the robot’s place. In 2D environments, students may experience spatial difficulties when planning rotations of objects by the robot. In contrast, interactive simulations in 3D virtual environments without immersion have been found effective [35–36] and are used in many robotics laboratories. Modern virtual robotic environments, such as RoboCell [37] and RobCAD [38], enable the learner to setup robotic cells and develop simulations of production processes. The virtual robotic cells can be made realistic and create some sense of immersion [39] by displaying simulated machinery, furniture, and other objects. Although different approaches to training spatial skills in STEM education have been widely discussed, very little research has considered spatial learning through practice in robotic environments. While studies relating spatial skills to robotics exist, most of these consider spatial ability skills only as prerequisites and predictors of learning achievement—e.g., Gomer and Pagano’s [40] study of advanced robotics courses for manufacturing engineering majors and Liu et al.’s [41] study of professional training courses for robot operators. Another example is Tseng and Yang [42], which will be discussed further below. In consequence, among 217 studies of spatial training in STEM education

Igor M. Verner and Sergei Gamer

reviewed by Uttal et al. [28], only two concerned robotics courses: the work in our laboratory [43] and the study of Gibbon [44]. We first encountered indications of spatial learning when guiding projects in which high school students programmed the instructional robot manipulator to automatically assemble block puzzles [45]. Our next paper [46] proposed a strategy for incorporating spatial training into the teaching of robot manipulator operation. Under the proposed strategy, each of the three main aspects of robotic manipulation—point-to-point control, rotation of objects, and robotic assembly—is used to strengthen one of the three main categories of spatial skills [16]: spatial perception, mental rotation, and visualization. In point-to-point control tasks, students observe various positions of the mechanical arm in the workspace and describe them analytically by means of coordinates. In rotation tasks, they design robot sequences to position blocks in desired orientations. In assembly design tasks, students visualize possible robot manipulations and identify those suitable to assemble the setup from the parts. Our subsequent article [43] implemented the proposed strategy in two middle school classes (grades 7 and 8) and one high school class (grade 10). Pre-course and post-course tests showed that the students made significant progress in tasks related to the spatial ability categories practiced in the course. Gibbon [44], in his case study, found that fifthand sixth-graders who engaged in construction activities with the LEGO Mindstorms Robotics Invention kit made significant gains in problem solving and spatial skills, as measured by the Raven’s Progressive Matrices test. This finding is in line with observations of [47–48]. Two additional studies involving spatial learning in the context of robotics are worth mentioning here. First, Tseng and Yang [42] taught a design course for sophomore mechanical engineering students at MIT. The course focused on the design of mobile robots and included spatial practice in origami making. The authors explored correlations between the students’ spatial abilities and the complexity of the robot design solutions in their projects. The findings showed that the students with higher spatial abilities at the start of the course showed better understanding of complex mechanical systems. Second, [49] report on the results of a training program for robotic surgery. The authors developed and implemented exercises to train surgeons in a set of primary skills, most of them related to robot manipulation and requiring spatial ability. Their preliminary findings showed that experts performed the exercises significantly better than novices,

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

underscoring the need for novices to develop the primary skills for robotic surgery. We note that both the mechanical engineering course [42] and the robotic surgery program [49] take as their starting point robot operation tasks that students need to master in order to perform the course assignments. In the study reported here, we start from the premise that spatial learning can be combined with teaching novices the principles of robot operation. We thus combine robotics studies (specifically, the operation of robot manipulators) with training spatial ability skills. When planning the tasks for the two case studies, we took into account the recommendations highlighted by [50]. The authors point to four aspects of spatial reasoning in interactive learning environments that are relevant for planning activities in robotics laboratories: 1.

2.

3.

4.

The picture superiority effect is the idea that people are better able to grasp and remember large amounts of information when that information is presented visually rather than verbally. In light of this effect, we designed the two courses so that each topic is introduced starting with a tangible demonstration of the robot operation. The demonstration facilitates further learning activities in the robotic environment, either physical or virtual. The noticing effect refers to the idea that people perceive more visual information in familiar compared with unfamiliar situations. This effect is used in robotics when practice with real robots follows training in a simulated virtual environment. In the current study, we therefore designed the course activities, where possible, so that students are introduced to each spatial task in a virtual environment, and only then work in real and remote environments. The structuring effect is an ability to track dynamic objects and extract their invariant structures. In operating robot manipulators, students need to be able to analyze and determine positions of the mechanical arm in the workspace. Lindgren and Schwartz [50] note that the structuring ability can be enhanced through practice in real and virtual environments. Our laboratory workshops include robotic tasks in which students track changes in the orientation of objects during robot manipulations. Based on the structuring effect, we argue that this practice can facilitate spatial learning and foster the development of structuring and mental rotation skills. The tuning effect is the improvement in rates of adaptation to new spatial manipulation tasks which comes with experience. The visual-motor

1379

calibration for performing a new robot manipulation task involves a number of skills: positioning the mechanical arm, perceiving the boundaries of the robot workspace, proficiency in robot motion control, and troubleshooting failures. In the laboratory, students develop their manipulation skills with a specific robot. As they practice with different workstations, they then become better at adapting their skills to robots of different types. Following Lindgren and Schwartz [50], the effects apply in virtual and physical environments. Moreover, the tuning and noticing effects, acting in combination, can provide transfer of skills between robotic environments of different types. Therefore, we can expect that training in a virtual environment will foster the development of robot operation skills in the real environment.

3. The Technion Robotics and CIM Laboratory The RCIM Laboratory in the Technion’s Faculty of Industrial Engineering and Management supports courses and activities for industrial engineering majors by enabling hands-on experimentation in the design, control and operation of automated manufacturing systems (Fig. 1). It also serves as a test bed for research experimentation by faculty members and graduate students. The laboratory facilities include a semi-industrial computer integrated manufacturing (CIM) system; computer aided design (CAD) workstations; a miniature flexible machining system (FMS) cell; a robot farm with five semi-industrial robots; and mobile robots. In terms of software, the lab is equipped with the RoboCell [37]. The CIM system in the foreground of Fig. 1 includes four robot manipulators and an automated storage-retrieving system (ASRS) connected by a conveyor loop. The entire system serves mainly for demonstration of robotic assembly of work parts supplied by the ASRS using the conveyor. In the Engineering Manufacturing Systems course for second-year students majoring in industrial engineering the robots are used also as standalone stations, though their workspaces for manipulations are limited. Similar RCIM laboratories, traditionally oriented toward advanced engineering courses, exist in other universities [11]. The CDIO approach requires substantial reorganization of such laboratories, turning them into modular workspaces for all levels of engineering education. For the current case studies, we modified the laboratory in keeping with the requirements of hands-on, virtual, and remote learning environments. Specifically, we customized the robot work-

1380

Igor M. Verner and Sergei Gamer

Fig. 1. Robotics and CIM Laboratory.

spaces; extended the RoboCell virtual environment; and enabled remote control of the robots through the local network. Several modifications were designed to enable students to perform robot manipulation tasks using oriented objects, such as cubes with signs and pictures on their faces. In these tasks, students must pick up each object from the storage area, rotate it to the desired orientation, and place it in the destination position. Using oriented objects significantly raises the spatial complexity of robot manipulations. The Faculty of Industrial Engineering and Management is interested in extending educational activities in the RCIM laboratory, and supported the required changes. The laboratory setup, including modifications, is described below. 3.1 Customizing the robot workspaces While in the regular laboratory setup robots are integrated into larger automated systems, in introductory courses they are operated as standalone stations. Each robot has an individual workspace suitable for its specific location and kinematic scheme. Fig. 2 shows a regular and modified robot setup. In Fig. 2A (the regular setup), the presence of system-related equipment (buffers, jigs, conveyer belts, etc.) limits the robot’s reachable workspace

and the number of possible manipulations. For the introductory courses, we constructed and installed in each workspace special superstructures that enable performance of the operation tasks (see Fig. 2B). Some of the robots in the laboratory are SCARA robots that do not have gripper pitch. To enable rotation manipulations using these robots, we constructed a LEGO rotator that can rotate objects (blocks) around horizontal axes. When the robot puts the block down on the rotator, the block slides into the predetermined position and simultaneously undergoes a 90o rotation. We also supplied plastic plates (pushers). The robot takes the pusher from the holder and uses it to align objects in the assembly area. Fig. 3A. shows the workspace of the robot SCORA-ER 14. Figs 3B and 3C present close-up views of the rotator and pusher. 3.2 Extending the RoboCell virtual environment to include oriented blocks. RoboCell is a software environment developed by Intelitek [37] that can be used to set up a virtual robotic workcell and create various subroutines for performing robot handling processes. This software makes it possible to ‘‘equip’’ virtual workcells with various items for learning and practice in robot programming and operation. These include the robot manipulator Scorbot ER5, CNC machines, part feeders, conveyors, work tables, welding tools, templates, jigs, and materials, and even items such as computers, pot plants, and chairs. Robot manipulations in virtual workcells created with RoboCell can be performed with parts having the shape of cylinders, cubes, and blocks. To enhance spatial learning in the virtual environment, we contacted Intelitek and asked them to modify RoboCell so as to enable students to define oriented cubes and manipulate them in the workcell. The

Fig. 2. Robot setup: A. Limited by four buffers and the conveyor; B. Adapted for performing manipulations.

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

1381

Fig. 3. A. SCORA 14 workspace for manipulation tasks; B. Cube rotator; C. Aligning tool (pusher).

company accepted our proposal and updated the software. This enabled us to offer virtual robot operation tasks in which students not only manipulate blocks of different lengths (Fig. 4A), but also rotate and orient cubes with different symbols on their faces (Fig. 4B).

3.3 Implementing remote control of the robots through the local network For each robot, we installed two 5M digital cameras and set them up to stream video footage of manipulations in the workspace to the local computer (Fig. 5A). Special ‘‘ACL Downloader’’ soft-

Fig. 4. A. Assembling blocks of different lengths; B. Manipulating cubes with symbols on their faces.

Fig. 5. A. Two cameras monitor the robot workspace; B. The screen view of the remote operator.

1382

ware, developed in the laboratory, was used to provide RS-232 communication between the computers and robot controllers. To control the robots, commands in the Advanced Control Language (ACL) developed by Intelitek [51] are entered into the command line of the Downloader. Feedback from the robot controller is presented in the Downloader control window, while the video streams are displayed in two special windows (Fig. 5B). Remote access was provided using the local computer network and the Windows Remote Desktop utility.

4. Educational study framework 4.1 Goal and questions The current study explores how different environments (hands-on, virtual, and remote) can be combined to benefit novice engineering students during practice in operating robot manipulators, in terms of both their understanding of robotics in general and their spatial skills in particular. Two case studies were conducted, each addressing one of two research questions: 1.

2.

Will laboratory practice in operating robot manipulators, based on the proposed approach, improve the performance of novice engineering students in spatial tasks, and if so, how? Will the practice in operating robot manipulators improve the students’ awareness of the importance of spatial skills in industrial robotics?

4.2 Participants The study involved two categories of participants. Participants in the first case study comprised 10th grade students majoring in mechanical engineering in an underprivileged vocational high school located in a Haifa suburb. Twenty students were studied over three years (two students in the first year, six in the second, and twelve in the third). Most of the students were regarded as academically disadvantaged. The course was designed at the request of the school to help students who were having spatial difficulties mastering technical drawing. Together with the technical drawing teacher, we developed an outreach robotics course to reinforce the students’ learning. In the second study, participants were first-year students of the Faculty of Industrial Engineering and Management. The faculty got interested in our robotics course for school students and asked to design on the same basis a robotics unit for the firstyear course ‘‘Introduction to Industrial Engineering and Systems Integration’’. Two hundred and forty-

Igor M. Verner and Sergei Gamer

eight students took part in the unit over three years (111 in the first year, 44 in the second, and 93 in the third). 4.3 Method Both case studies applied the action research method [52]. That is, the course and the unit were delivered several times, and each learning intervention was followed by (1) data collection, analysis, and interpretation; (2) adjustment of the learning environment and instructional strategies; and (3) implementation and evaluation of adjustments in the subsequent intervention. Thus, in the course of the case studies, we continually evaluated and improved the learning environment and instructional strategies employed. For the high school outreach course study, the assessed criterion was improvement in the students’ spatial skills, as measured by performance in robot operation tasks and in spatial tests. For the Technion course study, the criterion was the impact of the learning practice on students’ interest in industrial robotics and awareness of the importance of spatial skills in robot programming and operation. This criterion was measured using a perceptions questionnaire.

5. Case study 1 5.1 School outreach course The course was initially given in the 2011–2012 academic year (6 hours, two students), and was subsequently expanded in 2012–2013 (16 hours, six students) and 2013–2014 (16 hours, twelve students). As described above, all the students who took part were 10th graders learning technical drawing. Two school teachers helped adapt the course to the students’ needs. The teachers also administered the pre-course and post-course spatial tests for all students in the class, providing us with the opportunity to compare results of the students who participated in our course (the experimental group) and other students in the class (the control group). The 16-hour course, as delivered in 2013–2014, consisted of eight two-hour sessions. The curriculum was divided into three parts, where each part focused on a certain aspect of robot programming and operation, and on training one of the main categories of spatial ability: spatial perception, mental rotation, and visualization. The first three sessions focused on robot pickand-place operations in the workspace and spatial perception tasks. In the first session, after demonstration of an automated manufacturing process, the students learned about the structure of the robotic arm and its motion in the workspace. In

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

the second and third sessions they studied the robot control language ACL, learned to define robot positions by coordinates, and practiced programming simple pick-and-place manipulations with cubic parts. The second part of the course (three sessions) dealt with rotation of objects by the robot. In the fourth session the students learned about rotations around coordinate axes and how to perform them using the robotic arm. In the fifth and sixth sessions, they learned to use the RoboCell software and to operate a robot in the virtual environment. They completed this module by assembling a six-cube picture puzzle from identical cubes with geometrical symbols drawn on their sides (Fig. 4B). The seventh and eighth sessions (the third part of the course) were devoted to performing three assembly tasks with real robots. The first task was to assemble a six-cube picture puzzle through teleoperation, manipulating the robotic arm based on visual feedback from two digital cameras. The second task aimed to make a direct connection between our robotics course and the technical drawing curriculum that the students studied at school. For this task, the students were required to assemble a puzzle from six identical cubes with geometric figures drawn on their sides. The puzzle was presented using three orthographic projections (front, top, and side views) and a sketch (Fig. 6A). The students were asked to first use the sketch to depict a three-dimensional view of the puzzle by drawing appropriate geometric symbols on the sides. They then had to assemble the puzzle using the robot. The third task involved a Soma Cube puzzle, in which a 333 cube is dissected into block parts made up of smaller unit cubes joined at their faces [53]. Such block parts are highly suitable for assem-

1383

bling by a mechanical arm, and they allow a wide spectrum of spatial training activities. The students were required to use the block parts to assemble particular shapes (Fig. 6B). To do so, they had to design an assembly plan, determine the destination positions of each block, and finally, operate pickand-place manipulations. 5.2 Course development The version of the course described above represents the final iteration, after two years of development. The initial version of the course (2011–2012) consisted of three two-hour lessons, in which an introduction to robotics (first lesson) was followed by laboratory practice with real robots in lesson two (defining positions by coordinates and programming simple pick-and-place manipulations with cubic parts) and with the RoboCell virtual robotic environment in lesson three (programming and operating a virtual robot to assemble a structure from identical cubes with different symbols drawn on their faces). In analyzing the results of the course together with the technical drawing teacher, we saw the need to extend it so as to enable systematic training of spatial perception, mental rotation, and visualization skills. With this in mind, we revised the course for the second year (2012-2013) to include 16 hours, as follows:  2h lecture (introduction to robotics).  4h laboratory practice with the real robots (defining positions and programming pick-and-place manipulations with cubic parts).  2h lesson on the principles of rotation operations using the robot.  4h practice in the RoboCell virtual environment (assembling a structure from identical cubes with different symbols drawn on their faces).

Fig. 6. A. Geometric puzzle; B. Block parts and given shapes.

1384

 2h laboratory practice in robot teleoperation (assembling a picture puzzle).  2h laboratory practice in online robot programming using a teach pendant (a handheld control and programming device). The second year follow-up showed that the majority of students participating improved their spatial test scores, based on the spatial tests we administered (described below). In addition, all the students achieved high grades on the matriculation exam in technical drawing. Students’ own reflections on the course, as reported informally to us and to their teachers, were positive. For the third (final) version of the course, it was revised slightly to sharpen the connection with the technical drawing curriculum and motivate students to improve their spatial skills. In addition, it was decided to include more students from the technical drawing class. The following sections present the data collection procedures and findings from the final course. 5.3 Data collection At the beginning of each course we evaluated students’ spatial skills using three paper-andpencil spatial tests: the spatial perception test [54, p. 18], the mental rotation test [54, p. 290], and the visualization test [54, p. 149]. The spatial perception test requires students to reproduce a given pattern on a dot matrix. The test includes 32 patterns to be completed in 3 minutes, and is scored as the number of correct reproductions. The mental rotation test calls for comparing a pair of cubes with letters on their faces. Students have 3 minutes to solve 21 tasks. The score on this test is the number of answers marked correctly minus the number marked incorrectly. The three tests were repeated at the end of the course. In addition, we ran an interim spatial perception test at the end of the first part of the course and a mental rotation test at the end of the second part. The purpose of the interim tests was to provide feedback for lesson planning and to encourage students’ interest in the course. 5.4 Findings The results of the spatial tests for the third year show that the students in the course improved significantly both in relation to their initial scores, and in comparison to their classmates who did not take the course (the control group). Specifically, scores for the experimental group rose by 19.6% in the spatial perception test, by 104.5% in the mental rotation test, and by 30.1% in the visualization test compared with their pre-test scores. With respect to the comparison with the control

Igor M. Verner and Sergei Gamer

group, the students in the experimental group achieved higher average grades in the 2013 matriculation exam in technical drawing (88.0) compared with their classmates from the control group (83.3). The pre-course tests showed no significant differences in spatial performance between the experimental and control groups.

6. Case study 2 6.1 Freshman robotics workshop The workshop was delivered three times (in 2011– 2012, 2013–2014, and 2014–2015) as part of the firstyear course ‘‘Introduction to Industrial Engineering and Systems Integration’’. In the first year, the workshop took the form of a one-hour introduction to robotics and a two-hour laboratory exercise delivered by a teaching assistant. The exercise was given in the RoboCell virtual robotic environment [55]. The students were assigned to program a fivedegrees-of-freedom robot to assemble a structure consisting of different blocks. Based on results of the pilot, the Faculty dean challenged us to upgrade the workshop and provide experience in virtual and real robot operation for every first-year student. To meet this challenge, we developed a learning environment for practice in the programming and operation of virtual and real robots (described in Section 3) and a six-hour workshop curriculum. This curriculum included a lecture and two robotics laboratory classes. The lecture, ‘‘Principles of Robot System Operation,’’ introduced the students to the concepts of computerintegrated manufacturing, robot programming, and robot operation. In the second part of the lecture we presented the laboratory assignments. The first task for two-hour practice in the RoboCell virtual environment was similar to that given in the pilot workshop. In the second two-hour task, the students (in groups of two or three) operated real robot workcells that included a robot, a storage area for cubic parts, a buffer, and an assembly area. The parts comprised three identical cubes with digits from 1 to 6 irregularly oriented on their faces (Fig. 3C). In the assignment, the students were given two three-digit sequences that determined the final orientations of the cubes. For example, the sequence 612 determined the orientation presented in Fig. 7. The assignment required students to operate the robot so as to pick up each cube, move it from the storage area to the buffer, rotate it to the final orientation, and place it in the destination position at the assembly area. In this exercise, we aimed to free the students from the need to define different positions of the mechanical arm by means of coordinates, putting the focus instead on practice in planning and operating robot

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

1385

robots. A few students had studied robotics as an optional subject at school, and only one respondent had experience of work in a robotics company. Regarding the contribution of the workshop, 93% of the respondents reported that it exposed them to industrial robotics, and 17% evaluated this contribution as strong. Repeated reflections include: Fig. 7. Three cubes in the given orientation (up view).

movements in the workspace. For this purpose we developed subroutines, written in ACL, that implement basic pick-and-place operations. To operate the robot the students called the subroutines, one by one, using the ACL Downloader. The names of the subroutines are ‘‘spatial codes’’ describing the robot manipulations. For example, the subroutine XZ1 tells the robot to pick up a cube from the buffer with the gripper oriented horizontally (in the X-axis direction), move the cube upwards, turn the gripper to the vertical (Z) position, rotate it counterclockwise through angle  = 90o, and place it back on the buffer. The spatial codes were proposed by Verner [43]. That study found that the ‘‘pseudo-language’’ of spatial codes helps students plan and operate robot manipulations. Learning outcomes of the upgraded workshop delivered in the 2013–2014 course were evaluated through analysis of students’ reports and reflections. Based on this evaluation, in the 2014–2015 lecture we emphasized the role of industrial engineers in automated manufacturing. We posed the same robotic assignments for the laboratory classes, but explained them in more detail. 6.2 Students’ feedback After each workshop we administered a questionnaire to assess how the workshop influenced students’ understanding of the role of robotics in modern manufacturing, and their interest in the subject. Based on students’ responses during the first two years, we refined the questionnaire to be used at the next workshop. The final version included questions about students’ past experience in robotics and learning outcomes of the workshop. In particular, we asked the students if the workshop exposed them to industrial robotics and to the roles of industrial engineers in planning robotic production processes. We inquired about students’ interest in learning robotics and industrial engineering and their perceptions of the laboratory practice with virtual and real robots. Eighty of the 93 participants in the 2014–2015 workshop responded to the questionnaire. An absolute majority (92% of those who responded) had never studied robotics and had no experience with

This is the first time I was introduced to industrial robotics. The topic was very interesting and mainly exposed me to system optimization. The lecture and laboratory classes were fun and broke the routine. The lab tasks helped me see robots’ work more tangibly and understand the value of this work, especially of the new 21st-century robots designed to cooperate with people as partners in factories, classrooms, at home, etc.

More than half the respondents (65%) reported that the workshop effectively presented problems in operating and programming industrial robots; 23% considered this contribution to be high. Students noted: I thought this is an electrical engineer’s job. I was pleased to know that we (industrial engineers) have a part in it. I learned that industrial engineers working in robotic manufacturing explore robotic systems and technologies in specific work environments, using operations research, cognitive theories and optimization. I was very interested in robot operation planning and I would love to work on this in the future.

The workshop aroused students’ interest in studying robotics (55%), with about a quarter of the respondents reporting strong interest. From students’ reflections: I became interested in learning about robot capabilities and ways in which industrial engineers can use robots to optimize production and service processes. I want to study the implications of robotization on human resources and training. I was interested to learn the principles of robot operation and how different industrial robots work in cooperation. It is interesting to think about the extent to which human actions can be replaced by robots. I was interested in how to operate the robot in the most efficient way. I am also very interested in robot operation planning and programming. It was interesting to understand thinking required to operate the robot and perform the tasks successfully.

In addition to evaluation of the workshop’s impact, the questionnaire solicited students’ reflections on their practice in the virtual and physical robotic environments, including the difficulties they experienced in performing the relevant spatial tasks. Students’ evaluations of the practice in the virtual environment included the following: The virtual lab lets you perform operations with the robot without fear that something will break or go wrong. The virtual lab made it easier to understand considera-

1386

tions in planning robot operations: calculating angles, heights and positions. The virtual lab helped me understand the behavior of the robot, the principles of its operation and data processing. It is a good practice in planning manipulations in the workspace and enhances spatial vision.

Most of the students reported that they enjoyed their experience with real robots. From the reflections: The lab contributed greatly to my understanding of robot operation; it was fun to control the robot. The physical lab was much more interesting since it was a new work environment. The challenge was to think how to accomplish the task in the most effective way. It is hard to imagine robot operation without seeing how it is performed. I think we need to practice it because not everyone has good spatial skills.

In response to our request to compare the contributions of the virtual and physical labs, the students did not strongly favor one over the other. Rather, their responses suggest that both platforms serve important functions: In the virtual lab it is easier to understand the thinking behind operating the robot, calculating angles, heights and locations. The virtual laboratory helped me understand the principles of robot operation and the data entry process. It was fun to plan the assembly task with a minimum of blocks. The physical lab better demonstrates the robot workspace and gives an idea of the production process. It was a thrill to see in the physical lab how the program is carried out command after command.

The questionnaire asked the students to describe the spatial difficulties they experienced when performing the laboratory tasks. Some students reported difficulties in determining positions of the robot and rotation angles of the gripper (roll) in the assembly tasks; in taking account of limitations of robot mobility; and in assigning tolerances. Other difficulties were caused by inaccuracies in simulating the robot’s movements in the workspace.

7. Conclusion In this paper we presented our experience adapting the Technion Robotics and Computer Integrated Manufacturing Laboratory for introductory engineering courses. Using the CDIO (conceive-designimplement-operate) approach, we engaged first year IEM students in robotics activities and opened the laboratory to high school students majoring in mechanical engineering. Building on the educational opportunities afforded by human-in-theloop robotics, we focus the learning of novice students so as to help them understand the principles of robot operation, foster spatial skills and

Igor M. Verner and Sergei Gamer

awareness of their importance in industrial robotics. The key features of our approach are:  Customizing the robot workspaces to enable performance of spatial operation tasks.  Combining practice in direct, virtual and remote robot operation.  Extending the virtual robot environment to enable the manipulation of oriented blocks.  Directing robot operation tasks to train spatial perception, mental rotation, and visualization skills.  Delivering a lecture on the principles of robot operation in computer-integrated manufacturing. In our study, we sought to examine whether the developed approach can measurably improve the performance in spatial tasks and arouse the awareness of the importance of spatial skills in industrial robotics. We explored the developed approach for high school students majoring in mechanical engineering, and first-year IEM students. In the case of school students we assessed the performance of spatial tasks before and after our 16-hours course. As found, the students in the course improved significantly in perception, mental rotation, and visualization tests, both in relation to their initial scores and in comparison to their classmates who did not take the course. In the case of IEM students we assessed the impact of our 6-hours workshop using a perceptions questionnaire. The responses indicated that the workshop provided the first-hand experience in operation of real and virtual robots, helped to understand the emerging spatial problems and recognize the skills needed to cope with them. Based on the results of our study, obtained under specific conditions, we argue for further exploration of the proposed approach in different settings. As introductory robotics courses have become increasingly popular, we call for using them for training spatial skills that are highly demanded in engineering education and practice. Acknowledgements—This study was supported by Technion Gordon Center for Systems Engineering and PTC1 grants. The authors thank Prof. Avraham Shtub for help in the initial stages of the study and Prof. Sheryl Sorby for feedback on an earlier version of the article. We appreciate help of Technion and school instructors: Dr. Assaf Avrahami, Niv Krayner, Elena Baskin and Ronny Magril.

References 1. L. Strawderman and L. Ruff, Designing introductory industrial engineering courses to improve student career efficacy, International Journal of Engineering Education, 27(5), 2011, pp. 1019–1026. 2. K. K. Tan, L. J. Genalo and I. M. Verner, Guest editorial: Special issue on outreach to prospective electrical, electronics, and computer engineering students, IEEE Transactions on Education, 2010, 53 (1), pp. 1–2.

Robotics Laboratory Classes for Spatial Training of Novice Engineering Students

3. E. Crawley, J. Malmqvist, S. Ostlund and D. Brodeur, Rethinking Engineering Education: The CDIO Approach, Springer Press, New York, 2007. 4. D. Munro, Development of an automated manufacturing course with lab for undergraduates, IEEE Frontiers in Education Conference, 2013, pp. 496–501. 5. R. Pitts and R. Camilo, Integration and improvement of a robotics laboratory in an industrial engineering curriculum, Proceedings of the ASEE Annual Conference, 2010. 6. Z. Z. Bien and H.-E. Lee, Effective learning system techniques for human–robot interaction in service environment, Knowledge-Based Systems, 20(5), 2007, pp. 439–456. 7. I. M. Verner and S. Gamer, Focusing learner-in-the-loop robotics on training spatial skills. Proceedings of the IEEE International Symposium on Robot and Human Interactive Communication (RO-MAN), Edinburgh, Scotland, UK, 2014, pp. 672–677. 8. I. Ilyas and T. Semiawan, Production-based education (PBE): The future perspective of education on manufacturing excellence, Procedia- Social and Behavioral Sciences, 52, 2012, pp. 5–14. 9. 9. G. A. Chang and W. L. Stone, An effective learning approach for industrial robot programming, ASEE Annual Conference and Exposition, Atlanta, US, 2013. 10. J. Ma and J. V. Nickerson, Hands-on, simulated, and remote laboratories: A comparative literature review, ACM Computing Surveys, 38(3), 2006, pp. 1–24. 11. O. Goldstain, I. Ben-Gal and Y. Bukchin, Evaluation of telerobotic interface components for teaching robot operation, IEEE Transactions on Learning Technologies, 4(4), 2011, pp. 365–376. 12. C. S. Tzafestas, N. Palaiologou and M. Alifragis, Virtual and remote robotic laboratory: comparative experimental evaluation, IEEE Transactions on Education, 49(3), 2006, pp. 360–369. 13. C. A Jara, F. A. Candelas, S. T. Puente and F. Torres, Handson experiences of undergraduate students in automatics and robotics using a virtual and remote laboratory, Computers and Education, 57(4), 2011, pp. 2451–2461. 14. M. Stefanovic, The objectives, architectures and effects of distance learning laboratories for industrial engineering education, Computers and Education, 69(11), 2013, pp. 250–262. 15. J. E. Corter, J. V. Nickerson, S. K. Esche, C. Chassapis, S. Im and J. Ma, Constructing reality: A study of remote, hands-on and simulated laboratories, ACM Transactions on ComputerHuman Interaction, 14(2), 2007, pp. 1–27. 16. D. Lohman, Spatial abilities as traits, processes, and knowledge, Advances in the Psychology of Human Intelligence, 4, Lawrence Erlbaum Associates, Hillsdale, NJ, 1988, pp. 181– 248. 17. J. Wai, D. Lubinski and C. P. Benbow, Spatial ability for STEM domains: Aligning over 50 years of cumulative psychological knowledge solidifies its importance, Journal of Educational Psychology, 101(4), 2009, pp. 817–835. 18. D. Shin, R. A. Wysk and L. Rothrock, A formal controltheoretic model of human-automation interactive manufacturing system control, International Journal of Production Research, 44(20), 2006, pp. 4273–4295. 19. C. E. Lathan and M. Tracey, The effects of operator spatial perception and sensory feedback on human-robot teleoperation performance, Presence, 11(4), 2002, pp. 368–377. 20. M. A. Menchaca-Brandan, A. M. Liu, C. M. Oman and A. Natapoff, Influence of perspective-taking and mental rotation abilities in space teleoperation. Proceedings of the 2nd ACM/IEEE International Conference on Human-Robot Interaction, ACM, 2007, pp. 271–278. 21. S. Jones and G. Burnett, Spatial ability and learning to program. Human Technology: An Interdisciplinary Journal on Humans in ICT Environments, 4(1), 2008, pp. 47–61. 22. K. M. Stanney, J. Cohn, L. Milham, K. Hale, R. Darken and J. Sullivan, Deriving training strategies for spatial knowledge acquisition from behavioral, cognitive, and neural foundations, Military Psychology, 25(3), 2013, pp. 191–205. 23. P. Peruch, L. Belingard and C. Thinus-Blanc, Transfer of

24.

25.

26. 27. 28.

29. 30. 31. 32. 33.

34.

35. 36.

37. 38. 39. 40.

41.

42.

43.

1387

spatial knowledge from virtual to real environments, Lecture Notes in Artificial Intelligence, Spatial Cognition II, 2000, pp. 253–264. B. Ben Youssef and B. Berry, Learning to think spatially in an undergraduate interdisciplinary computational design context: A case study, International Journal of Technology and Design Education, 22(4), 2012, pp. 541–564. Z. Merchant, E. T. Goetz, W. Keeney-Kennicutt, O. M. Kwok, L. Cifuentes and T. J. Davis, The learner characteristics, features of desktop 3D virtual reality environments, and college chemistry instruction: A structural equation modeling analysis, Computer Education, 59(2), 2012, pp. 551–568. D. I. Miller and D. F. Halpern, Can spatial training improve long-term outcomes for gifted STEM undergraduates? Learning and Individual Differences, 26, 2013, pp. 141–152. J. L. Mohler, A review of spatial ability research, Engineering Design Graphics Journal, 72(2), 2008, pp. 19–30. D. H. Uttal, N. G. Meadow, E. Tipton, L. L. Hand, A. R. Alden, C. Warren and N. S. Newcombe, The malleability of spatial skills: A meta-analysis of training studies, Psychological Bulletin, 139(2), 2013, pp. 352–402. S. A. Sorby, Educational research in developing 3-D spatial skills for engineering students, International Journal of Science Education, 31(3), 2009, pp. 459–480. S. Hsi, M. Linn and J. Bell, The role of spatial reasoning and the design of spatial instruction, Journal of Engineering Education, 86(2), 1997, pp. 151–158. C. Diplas and V. Komis, Spatial abilities in virtual environment’s learning spaces, International Journal of Learning, 18(10), 2012, pp. 63–82. E. A. L. Lee and K. W. Wong, Learning with desktop virtual reality: Low spatial ability learners are more positively affected, Computers and Education, 79, 2014, pp. 49–58. M. Kozhevnikov and A. Garcia, Visual-spatial learning and training in collaborative design in virtual environments. In X. Wang & J. J.-H. Tsai (Eds.), Collaborative Design in Virtual Environments, ISCA, vol. 48, Springer Netherlands, 2011, pp. 17–26. R. L. Klatzky, Allocentric and egocentric spatial representations: Definitions, distinctions, and interconnections, In C. Freksa, C. Habel, and K. F. Wender (Eds.), Spatial cognition, Springer Berlin, Heidelberg, 1998, pp. 1–17. J. Trindade, C. Fiolhais and L. Almeida, Science learning in virtual environments: A descriptive study, British Journal of Educational Technology, 33(4), 2002, pp. 471–488. B. Caci, A. D’Amico and G. Chiazzese, Robotics and Virtual Worlds: An experiential learning lab, in A. Chella et al. (Eds.), Biologically Inspired Cognitive Architectures, AISC196, Springer Berlin Heidelberg, 2012, pp. 83–87. Intelitek, RoboCell for Controller USB, http://www. intelitekdownloads.com/Manuals/ Robots/ER-4u/100346G%20RoboCell-USB-v56.pdf, Accessed 22 March 2015. Siemens, Robcad, http://www.plm.automation.siemens. com/en_us/products/tecnomatix/ manufacturing-simulation/ robotics/robcad.shtml. Accessed 22 March 2015. N. Rutten, W. R. van Joolingen and J. T. van der Veen, The learning effects of computer simulations in science education, Computers and Education, 58(1), 2011, pp. 136–153. J. Gomer and C. Pagano, Spatial perception and robot operation: Should spatial abilities be considered when selecting robot operators? In Proceedings of the Human Factors and Ergonomics Society Annual Meeting, SAGE Publications, 55(1), 2011, pp. 1260–1264. A. M. Liu, C. M. Oman, R. Galvan and A. Natapoff, Predicting space telerobotic operator training performance from human spatial ability assessment, Acta Astronautica, 92(1), 2013, pp. 38–47. T. Tseng and M. Yang, The role of spatial-visual skills in a project-based engineering design course, Proceedings of the ASEE Annual Conference and Exposition, 2011, Vancouver, BC, Canada. I. M. Verner, Robot manipulations: A synergy of visualization, computation and action for spatial instruction, International Journal of Computers for Mathematical Learning, 9(2), 2004, pp. 213–234.

1388

44. L. Gibbon, Effects of LEGO Mindstorms on convergent and divergent problem solving and spatial abilities in fifth and sixth grade students. Ed.D. dissertation, 2007, Seattle Pacific University, Washington, http://search.proquest.com/ docview/304711988, Accessed 22 March 2015. 45. S. Waks and I. Verner, Positioning issues in the study of robotic manipulations, International Journal of Engineering Education, 9(3), 1993, pp. 223–230. 46. S. Waks and I. Verner, Spatial vision development through manipulating robot movements, European Journal of Engineering Education, 22(1), 1997, pp. 35–43. 47. S. V. Coxon, Challenging neglected spatially gifted students with FIRST LEGO League. Addendum to leading change in gifted education, Center for Gifted Education, Williamsburg, VA, 2009, http://cfge.wm.edu/Documents/FestschriftSupple ment.pdf#page=25, Accessed 22 March 2015. 48. S. V. Coxon, The malleability of spatial ability under treatment of a FIRST LEGO league-based robotics simulation, Journal for the Education of the Gifted, 35(3), 2012, pp. 291–316. 49. G. Dulan, R. V. Rege, D. C. Hogg, K. M. Gilberg-Fisher, N.

Igor M. Verner and Sergei Gamer

50. 51.

52. 53. 54. 55.

A. Arain, S. T. Tesfay and D. J. Scott, Developing a comprehensive, proficiency-based training program for robotic surgery, Surgery, 152(3), 2012, pp. 477–488. R. Lindgren and D. L. Schwartz, Spatial learning and computer simulations in science. International Journal of Science Education, 31(3), 2009, pp. 419–438. Intelitek, Advanced Control Language for Controller B, http://www.intelitekdownloads.com Manuals/Robots/ Obsolete/Controller_B/100085-a%20ACL28-Ctrl-B.pdf, Accessed 22 March 2015. B. Somekh, Action Research: A Methodology For Change and Development, McGraw-Hill International, 2005. S. Coffin, The Puzzling World of Polyhedral Dissections, Oxford University Press, Oxford, USA,1991. J. Eliot and I. Smith, An International Directory of Spatial Tests, NFER-Nelson, NJ, USA, 1983. I. M. Verner, S. Gamer and A. Shtub, Fostering students’ spatial skills through practice in operating and programming robotic cells, in J.-H. Kim et al. (eds.), Robot Intelligence Technology and Applications, Springer Berlin Heidelberg, 208, 2013, pp. 745–752.

Igor Verner is Associate Professor in the Department of Education in Science and Technology with secondary affiliation in the Faculty of Industrial Engineering and Management at the Technion—Israel Institute of Technology. His research interests include learning, teaching, and professional development through design and operation of robot systems. He is Head of the Center for Robotics and Digital Technology Education. He holds a M.S. degree in mathematics and a Ph.D. degree in computer aided design systems in manufacturing from the Ural Federal University, and a teaching certificate in technology from the Technion. Sergei Gamer is a Ph.D. student in the Department of Education in Science and Technology and Technical Manager of the Robotics and Computer Integrated Manufacturing Laboratory at the Faculty of Industrial Engineering and Management at the Technion—Israel Institute of Technology. His research interests include engineering education and development of spatial skills in robotic environments. He holds a M.S. degree in Electrical Engineering from the Belarusian National Technical University.