Virtual Creatures VWSIM98 Conference Paper

THE DESIGN OF FROG ISLAND: A VRML WORLD FOR BIOLOGY

Ramani Pichumani Ph.D., Decker Walker Ph.D., W. LeRoy Heinrichs M.D. Ph.D., Chandu Karadi Ph.D., William Anthony Lorie, Parvati Dev Ph.D.

Stanford University Medical Media and Instructional Technologies
Stanford University School of Medicine
Stanford, CA 94305
E-mail: ramani@smi.stanford.edu, dev@smi.stanford.edu

We have designed and implemented a prototype web-based virtual 3D environment for teaching vertebrate biology for high school and middle school students. This 3D learning environment, which we called Frog Island, contains a Virtual Frog along with a rich array of related resources (images, sounds, data, and simulations) that students and teachers can use to study the biology of frogs. Working closely with biology teachers and students, we developed a new Web-based approach for teaching biological principles: engage students in an interactive, engaging, self-paced, virtual environment which contains a rich array of multimedia learning resources. The virtual creature, and the tools to explore, dissect and reconstruct it, will complement existing techniques such as class lectures, textbooks, lab experiments, and dissections. Our objective is to improve the effectiveness of the biological and life sciences curriculum in secondary schools by introducing new computer-based learning methods that will allow students to interact with, experiment on, and perform goal-based exploratory and constructive exercises in virtual environments. Although initial reviews with teachers have been favorable and have shown this technique to be useful in classroom settings, we are in the process of conducting formal evaluations of our approach with a broader study group.

Recently, fewer and fewer students have been choosing careers in math, science, and engineering. More and more schools are having difficulty purchasing supplies and resources needed to teach physical and life sciences effectively. And yet, to prepare themselves to tackle many of the future problems in science and industry, students will need more highly developed problem-solving skills and more sophistication in managing, analyzing, and utilizing scientific information. Innovative computer technologies can be to used to develop highly engaging, problem-based teaching modules that will allow students to explore science in virtual laboratories and other information rich settings. Students will be able to interact with computer simulations of organic and inorganic systems within virtual environments that incorporate sophisticated models of the systems under study. Such environments should allow for both structured as well as unstructured (exploratory) modes of learning. Using Internet-based technologies, students will be able to share information among their peers, teachers, and the larger Web community.

The primary goal of the Virtual Creatures project at Stanford University is to improve the effectiveness of the biological and life sciences curriculum in secondary schools by introducing computer-based learning methods that allow students to interact with, experiment on, and perform goal-based simulations using a set of virtual 3D creatures. These new methods will complement existing techniques such as class lectures, textbooks, lab experiments, and dissections. They will enable students to learn, retain, and use the knowledge they have acquired in more effective ways than was previously possible. Our hypothesis has been that students will be able to learn and retain more factual knowledge, while at the same time enhancing their reasoning and problem-solving skills as they make use of the following computer-based teaching tools:

The Virtual Creatures project incorporates the collaborative efforts of specialists in education, biology, computer graphics, biomechanical engineering and education technology. We expect projects like this to help establish a benchmark of the type of research that is achievable only through such collaboration.

Working under a grant from the National Science Foundation, a multidisplinary group of researchers from the Stanford University School of Education and the School of Medicine initiated a project to use realistic 3D graphics and collaborative Web-based instruction to teach vertebrate biology. The original goal was to create a virtual creature that students could dissect on computer (see Figure 1). The virtual creature would consist of a 3D volumetric rendering derived from computer tomography of an actual creature. Students would be able to do simulated endoscopic examination as well as simulated surgery on the virtual creature. We developed a prototype virtual creature using a frog dataset (Johnston 1994), reviewed research on the teaching and learning of anatomy, and interviewed teachers and students to determine how to effectively this data.

**Figure 1:** A manipulable 3D Virtual Frog shown with various organ systems.
$\begin{figure} { \psfull \centerline{ \psfig {file=whole-frog2.eps,width=3.4in} } }\end{figure}$

What we learned in this work led us to greatly expand our original goal. The Virtual Creature that students could explore in various ways evolved into a simulated 3D learning environment, which we called Frog Island, and which contains the original Virtual Frog along with a rich array of related resources that students and teachers can use to study the biology of frogs - images, sounds, data, and simulations (http://summit.stanford.edu/creatures/frog-island.html). Working closely with biology teachers and students, we developed an approach which amounts to a new way of teaching biological principles: engage students in an interactive, engaging, self-paced, virtual environment which contains a rich array of multimedia learning resources.

Frog Island is an environment where students embark on an exploration of a 3D virtual outdoor habitat and visit various ``huts'' to learn about topics such as anatomy, physiology (digestion, respiration, circulation), biomechanics, and diversity (see Figure 2). They have access to a workbook containing an outline of the various lesson plans. This will allow them to enter observations, to raise questions, or to enter answers to questions that are posed. The workbook will help students stay focused on a particular lesson plan without getting lost during the course of visiting various links. In addition to the workbook and VRML windows, there is also a supplementary window where relevant text, images, and videos can be displayed. To facilitate unstructured learning modes, students are also encouraged to explore the virtual environment on their own, interacting and noting any observations they may care to make.

**Figure 2:** Scenes from the Frog Island 3D virtual world.
$\begin{figure} { \psfull \centerline{ \psfig {file=frog-island-pix.eps,width=3.4in} } }\end{figure}$

The virtual habitat resembles a natural setting with ponds, grass, trees, and meadows. There are various theme-based activity huts and regions that provide students with various resources needed to complete their lesson plans and activities. The first hut that students are guided to is the Orientation hut. Here they get an overview of the environment, the learning goals, and the lesson plans they will encounter. A friendly ranger or nature interpreter provides the introductory remarks.

We have designed and/or developed the following huts for our prototype Frog Island environment:

Students visit the orientation and diversity huts before embarking on the more challenging places such as the Biomechanics hut. Whenever they choose to visit huts to engage in activities that require them to be well grounded in fundamental knowledge (e.g., anatomy and physiology), they will be invited to explore the fundamental concepts before continuing. This will hopefully result in more powerful context-based learning activities within a goal-driven environment. In most of the huts, students can access in-depth information on specific topics by clicking on books on the reference shelf. Figure 3 shows a view from inside the Organs Systems hut where the internal organs of the virtual frog are revealed. By clicking on any organ on the frog, students can received in depth information about that organ in the lower window. Students can also access more general topics by clicking on the reference books which are organized by organ systems. Furthermore, students can access the entire frog information frames database by clicking on the computer screen next to the virtual frog. The information frames (see Figure 4) offer content at multiple levels from complex to simple topics. For the basic physiological and anatomic structures of the frog, we include the topics of bones, joints, muscles, nerves and circulation. For presenting the many facts about these anatomic structures and their functions, we have designed a standard organizational template: the topic covered by the frame is presented in the header, and the location (address) of the frame is given by the frame number. A collection of frames with its links to other topics and frames is referred to as a concept map.

One of the key features of Frog Island is the bridging of hypertext multimedia information frames and concept maps to the 3D virtual environment. This allows students to incorporate both structural as well as textual information in a single unified context. Furthermore, it allows for parallel development and use of information frames and 3D models, and promotes a linked representation of these two content modalities. The organization of the concept maps also allows students who lack VRML-capable browsers to access the same content in the information frames without the links to the 3D virtual objects.

**Figure 3:** The Organ Systems hut contains a 3D representation of the virtual frog which students can view from arbitrary angles, selectively include and exclude various organs, and access multimedia information linked to each organ (such as the liver shown above).
$\begin{figure} { \psfull \centerline{ \psfig {file=organshut.eps,width=3.4in} } }\end{figure}$

**Figure 4:** An information frame is a self-contained packet of information about a biological concept which is presented in a standard format. The organization of a frame allows students to quickly view related topics in greater depth without losing track of the conceptual relationships of those topics.
$\begin{figure} { \psfull \centerline{ \psfig {file=iframes.eps,width=3.4in} } }\end{figure}$

One of the design goals was to make it easy for teachers to add their own content or augment the existing content with their own resources. In addition, we have facilities to instrument the students activities (implicit record keeping) as well as allowing students to record and publish their own findings (explicit record keeping).

Frog Island has been designed to allow students to run simulations of frog behavior in the context of the 3D environment. There are three types of simulations we have targeted: (1) Java-based simulations of simple vertebrate physiology and behavior, (2) VRML-based simulation of skeletal movements, and (3) finite-element modeling (FEM) of organ systems and musculoskeletal movements. Due to the difficulties of implementing FEM techniques over the web, at present we have developed only the first two types of simulations.

For Java-based simulations, we have developed the Jumping Frog game and the Muscle Composition Lab. The Jumping Frog game is an interactive 2-D Java applet that teaches children about simple frog behavior (eating, jumping, survival) as well as simple projectile motion (see Figure 5). The goal of this game is to teach biology and physics in an integrative, interdisciplinary manner. There are two major goals of this game. The first mode is the skill development mode where the student learns about angular measures, weight, force, and parabolic trajectories. The student is taught how to make a frog jump in an optimal manner using a minimum amount of energy. In the process, the student develops an understanding of simple projectile motion. The second mode is the survival game in which the student uses the skills he/she has developed to help a frog eat and survive. The student wins points for eating flies (thereby increasing their energy store) while at the same time must avoid being eaten by predators (in this case, birds). When threatened by a bird, the student must make the frog jump into a pond. If the frogs misses the pond, it becomes disabled for a certain number of seconds before jumping again. If the frog eats too much food, it becomes too massive to jump efficiently. Likewise, if the frog eats too little, it becomes weak and unable to jump effectively.

**Figure 5:** The Jumping Frog game teaches simple frog behavior (eating, jumping, survival) as well as the principles of simple projectile motion.
$\begin{figure} { \psfull \centerline{ \psfig {file=jumpingfrog.eps,width=3.4in} } }\end{figure}$

The second Java-based simulation demonstrates how different muscles types have different strengths, reaction times, and endurances. In this Muscle Lab, students conduct two activities: (1) the muscle composition experiment, and (2) the race track game. Using a worksheet that guides them through the activity, students learn what combinations of muscle types would be ideal for (1) winning a jumping frog contest, (2) survival in the frog's natural habitat, (3) and for different types of athletes, what types types of muscles are best suited for different areas of the body (e.g., back muscles versus the biceps). In addition, some students will be interested in seeing the mathematical equations that comprise the muscle model to gain better understanding of their behavior and to be able to make predictions of their experimental outcomes. Once students have mastered the basic properties of different types of muscles, they can then play a multi-user track race game in which they select the muscle composition of their athlete. The goal of this simulation game (written in Java) is to win competitions including a 100 meter, 400 meter, and 1600 meter race (see Figure 7).

The second type of simulation uses features of the VRML 2.0 language (Hartman 1996) to demonstrate the motion of joints such as the knees and hips. Students can point and click on various joints to see how their movement relates to the function of the particular joint (see Figure 6). VRML allows for pre-scripted movements of objects and object hierarchies which can be activated directly by users, or indirectly through other operations. While this is suitable for certain types of rigid-body motion, simulating deformations of soft tissues is very problematic. Furthermore, VRML does not have general inverse kinematic capabilities which are needed for goal directed movement. As a result, the kinematics of the Virtual Frog is currently limited to skeletal movements.

**Figure 6:** In the Bones Hut, students can examine the various bones of the frog skeleton. Clicking on a joint will demonstrate how that joint moves and what function it provides.
$\begin{figure} { \psfull \centerline{ \psfig {file=frogbones.eps,width=3.4in} } }\end{figure}$

**Figure 7:** A multi-user race game simulation is used to test student's understanding of muscle physiology. Each student designs an athlete with various muscle composition ratios and watches them compete against other student's athletes in speed as well as endurance races.
$\begin{figure} { \psfull \centerline{ \psfig {file=javarace.eps,width=3.4in} } }\end{figure}$

Our initial studies with teachers and students indicated that there is strong interest in using Frog Island in the classroom. In order to determine how successfully students and teachers use the various resources provided in the current version of Frog Island, we are in the process of conducting broad studies with individual users, students and teachers, and focused, experimental studies in which users are exposed to different versions of the site. We also seek to assess their use of the site, their thinking, their achievement, and their motivation for further learning. In addition, we will evaluate the following questions:

Although Frog Island has been publicized and made accessible to all K-12 schools via the Web-site, we plan to limit the formal evaluation of its effectiveness to a sample size of 20-30 students and 10 teachers from local area schools. A larger evaluation will require significantly more resources than we have available at this time.

There are three areas of Frog Island that are targeted for future work: (1) expanded content, (2) authoring environments for teachers, and (3) biomechanical and kinematic modeling. We would like to expand the content of Frog Island to include more physiological information on frogs as well as other creatures such as humans and birds. This will allow for better comparative learning modes involving a larger spectrum of vertebrates. The goal of expanded content can be achieved by developing authoring environments that allow teachers to create their own content to customize and extend the number of teaching models. At present, VRML authoring tools require a significant amount of learning time. The vast majority of teachers do not have time to devote to learning computer-based authoring tools. A more viable approach would involve using templates, filters, and stock 3D models to create HTML and VRML pages that can be customized for particular learning activities. The most challenging goal at this time is the incorporation of FEM techniques to simulate biomechanical and kinematic behavior in a VRML environment. While there are many machine-dependent approaches that can be used, our goal with Frog Island is to provide accessibility over the Web in order to reduce the complexity of installing, running, and managing the software. At present, this goal is still elusive and may require writing machine-specific simulation modules that must be installed separately.

The Virtual Creatures project is aimed primarily at the development and deployment of powerful learning technologies for generating advanced software for students of all ages who are studying biology. Frog Island is the first implementation of a 3D virtual habitat for studying frogs in an interactive environment. At present, students can interactive with this environment using any VRML enabled browser. The major challenge ahead is to integrate a simulation engine that will allow for modeling of behavioral, physiological, and biomechanical properties. The current VRML standard does not provide for an efficient, machine-independent method for integrating powerful simulation capabilities into web browser platform. Nevertheless, our hope is that the Virtual creatures will, over time, become as universally available as textbooks, whether for classroom study, recreation, or life-long learning.

We are greatly indebted to William Johnston of the Information and Computing Sciences Division at Lawrence Berkeley National Laboratory for providing the original datasets used to create their dissectible frog on the Internet. The authors would like to acknowledge the contributions of David Braginsky, Ron Frankel, Chad Hollingsworth, Ashley Holt, Phil Li, Patrick Roisen, and Chris Thu, all of whom participated in the development of Frog Island. The Virtual Creatures project was supported in part by NSF grant CDA 9616441, DARPA CAETI grant N66001-95-C-8618-P0001 and by a matching State of California grant C-94-0022.

Dr. Pichumani heads the research efforts in visualization and modeling at the Stanford School of Medicine's SUMMIT Lab. He received a Ph.D. in Medical Informatics in 1996 from Stanford University and has eight years of industrial and academic experience in the field of three-dimensional image processing, visualization, and computational methodologies. His Ph.D. dissertation dealt with finite element model representations, optimization techniques, and model-based image segmentation algorithms for medical imaging.

Dr. Walker received his Ph.D.in Education from Stanford University, where he has been a faculty member since 1990. He has authored many books and papers ranging from formative studies in the design of a flexible computer classroom, design of interactive media with students, a guide for teachers to incorporate computer technology into their teaching, and ways to prevent the dehumanization of education by technology.

Dr. Heinrichs is Professor Emeritus and past-chair of Gynecology/Obstetrics at Stanford. He coordinates the development at SUMMIT of 3D models of human anatomy, creating Virtual Organs for teaching and simulated videoendoscopic surgery. Prof. Heinrichs has devoted his academic career to research into clinical methods (including the use of computer modeling) to improve surgical methods in endoscopic pelvic surgery.

Dr. Karadi holds a Ph.D. in Physics from the University of California, Berkeley, and is currently pursuing an M.D. at the Stanford Medical School. Dr. Karadi was chief architect of the Frog Island VRML environment.

Dr. Dev is the principal investigator of the Virtual Creatures project and is active in research and teaching in the field of medical imaging, visualization and modeling. From 1982 to 1989 she led product research and development for three-dimensional imaging of patients from computed tomography and magnetic resonance scans. Since then she has been the director of the SUMMIT Lab at Stanford University and is in charge of developing and encouraging the use of computers in medical education. Her research focuses on technologies to increase the efficiency and quality of education software development, and recently, on authoring tools for the construction of virtual teaching spaces on the Internet.

Arnold, Ken; Gosling, James. 1996. The Java Programming Language. Addison-Wesley, Reading, MA.

Greeno, James G; Goldman, Shelly V. (eds.). 1997. Thinking Practices. Lawrence Erlbaum Associates, Hillsdale, NJ.

Hartman, J. 1996. The VRML 2.0 Handbook: Building Moving Worlds on the Web. Addison-Wesley, Reading, MA.

Johnston, W.E. ``The Whole Frog Project.'' (Web page) http://george.lbl.gov/ITG.hm.pg.docs/Whole.Frog/-Whole.Frog.html. 1994. University of California, Lawrence Berkeley Laboratory, Berkeley, CA.

Peterson, Donald (ed.). 1996. Forms of Representation. Intellect Books, Exeter, England.

Pichumani, R.; Walker. D.; Karadi. C.; Heinrichs, W.L.; Lorie, W.L.; Dev, P. ``Virtual Creatures.'' (Web page) http://summit.stanford.edu/creatures/. 1997. Stanford University, Stanford, CA.

Robertson D.; Johnston, W.E.; Nip, W.; ``Virtual Frog Dissection: Interactive 3D Graphics Via the Web.'' In Proceedings of the Second International WWW Conference: Mosaic and the Web. 1994. Chicago, IL.

Schenk, Roger C. 1995. Engines for Education. Lawrence Erlbaum Associates, Hillsdale, NJ.

Schroeder, Will; Martin, Ken; Lorenson, Bill. 1995. Visualization Toolkit. Prentice-Hall, Englewood Cliffs, NJ.

Walker, F. Decker; Hess, Robert D. (eds.). Instructional Software : Principles And Perspectives for Design and Use. 1984. Wadsworth Pub. Co., Belmont, CA.