THE CUBE: A 3D Vibration Jigsaw Puzzle

Project team members: Colette Posse and Sami Shad

"THE CUBE" is a 3D jigsaw puzzle, with the added feature that a puzzle piece vibrates for approximately a second when it is placed in the correct place in the puzzle. We accomplished this by building an electrical design such that a circuit delivering power to a vibration motor was completed each time a puzzle piece was correctly placed onto a base. We hope that our project will reach the educational goal of improving the process of learning spatial skills by providing positive feedback and encouragement. Our qualitative results indicated that the vibration feedback did make the puzzle experience more fun: our table was occupied with visitors for the duration of the event, and "THE CUBE" was voted "Most Creative". When completed, "THE CUBE" looks like the picture to the right:

Introduction

In our final haptics project, we aimed to build a haptic device that targeted a more fundamental skill in life than something learned in a high school. We began considering concepts learned at a younger age than high school to meet this interest and came to the conclusion that we wanted to influence learning of spatial awareness in children. To explore the effects of haptic feedback on the experience of developing spatial awareness in children aged seven through ten, we came up with "THE CUBE".

Background

Below is a description of our research into previous work as it relates to several aspects of our project-- spatial learning, 3D puzzles, and vibrotactile feedback. Each of these sources gave us some sort of insight and/or molded our approach.

1) Tangible user interfaces: tools to examine, assess and treatdynamic constructional processes in children with developmental coordination disorders

Summary: This paper describes a set of cubes that, when interlocked, rendered their position and orientation onto a computer monitor. Children with and without Developmental Coordination Disorder (DCD) played challenging visual games with these blocks, and their spatial intelligence measurably improved.

Significance to our project: Although we came across this paper after we completed the project, it was still quite useful for future work purposes. This experimental procedure is similar to one we could implement if we were to test our puzzle on our target age group (7-10 years). With specific challenges and a more modular design, we could make spatial learning even more measureable.

2) Evaluation of a Vibrotactile Feedback Device for Spatial Guidance

Summary: This paper describes an experiment that compared spatial performance with vibrotactile cues to that with verbal cues. This was done by commanding the right arm to translate and rotate to different orientations. The result (concluded from the number of errors made in each case) was that vibrotactile feedback was more effective with rotational commands, and verbal feedback was more effective for translational feedback.

Significance to our project: We decided to go with a 3D puzzle instead of a planar one because of this paper. If pieces must be oriented correctly in 3D space, it will maximize the effect of vibrotactile feedback features of the puzzle.

3) Development of spatial memory and spatial orientation in preschoolers and primary school children

Summary: This paper gives results of a series of tests designed to measure the spatial competencies of 5, 7, and 10 year olds. The tests consisted of
1) Training: a normal maze with cues as to where to go and find certain objects.
2) Rotating the perspective of the user 180^o.
3) Rotating all the cues.
4) Deleting cues.
5) Deleting cues and then rotating the perspective of the user.
The results found that as the tests became increasingly more difficult, the younger children got more and more lost and made more errors. The main result of the paper is that the development period for complex spatial awareness and development was confirmed at the ages between 5 and 10, as previously suspected.

Significance to our project: This could point us to the target age group of our 3D puzzle for the sake of spatial skill development. If we market this toward early elementary school children, it will be most effective according to this study.

4) Virtual 3D Jigsaw Puzzles: Studying the Effect of Exploring Spatial Relations with Implicit Guidance

Summary: This paper talks about an experiment to test the spatial knowledge gained by physiotherapy students from solving virtual 3D puzzles related to human anatomy. The effectiveness of the puzzles was evaluated before and after the puzzles by having the students: 1) answer questions about anatomical facts 2) take a spatial intelligence test, 3) talk about their confidence in that anatomical knowledge, and 4) talk about the usability and difficulty with the puzzles. The study concluded that not only did the 3D puzzles improve anatomical and spatial knowledge and intelligence, but they also greatly increased motivation for learning such things.

Significance to our project: We became more confident in our choice of project after reading this paper, and broadened our possible audience to medical students trying to learn anatomy in a way that is not just rote memorization. This also got us thinking about anyone else that would need to have 3D spatial awareness for their job and might want to learn it in a more fun way—mechanics, electricians, plumbers, and interior designers were a few we came up with. 3D puzzles could be customized for training for each of these professions.

5) Sensory Puzzles, CHI '99 Extended Abstracts on Human Factors in Computing Systems

Summary: This paper is very brief, but it describes the design of a 3D puzzle for blind or visually impaired children which plays music for each piece successfully positioned. Each piece also has an interesting surface topology corresponding to the music it plays.

Significance to our project: We were hoping we would get some information about how this idea worked for the children that it was meant for, but this was too preliminary. We did gain from this the fact that our idea was novel in that it provided realtime haptic feedback. This was the most similar project we could find in existence.

6) Comparing 6DOF haptic interfaces for application in 3D assembly tasks

Summary: This paper discusses the use of virtual environments to solve 3D puzzles, for eventual use for surgical or archeological reconstruction. Subjects were asked to put together a 3D virtual puzzle with the Phantom Premium, Delta, and Spidar robots. It was found that with force feedback, task completion times decreased significantly. The Spidar was subjectively found to be the easiest to use because it allowed for more rotational freedom.

Significance to our project: Although we eventually rejected the idea of creating a virtual puzzle instead of a physical one, this paper got us thinking about that possibility. It would be a great future step once we have a physical puzzle made. This sort of design would be easy to change and manipulate for different purposes, however, it would probably be less fun to play with.

Methods

One of the unique features of "THE CUBE" is that all the logic it uses is determined by its hardware. This came about through our design requirement of running "THE CUBE" off just a power and ground connection. "THE CUBE" had to know when the connections were established and vibrate for a set amount of time after the connections were made.

Hardware Design and Implementation, System analysis/control

Electrical Design:
Our electrical design consisted of a motor that ran off of a 12 V power supply. The motor was connected to a MOSFET so turning the motor on and off was controlled by a logic circuit. We designed the logic circuit to run off of 5V, so used a voltage regulator to run the logic circuit. The circuit itself had a high pass filter, so when the connection was made, the input to the filter saw a 5V step input, and output a decay with an exponential decay to 2.5 V into the positive input of a comparator, with the negative input at 2.6V, so the transistor would be in its ON state for a short amount of time after the step input.
The motor we used had an eccentric rotating mass on its shaft, so caused the puzzle piece to vibrate whenever it was on. The schematic below shows our circuitry, with all the components in the box inside each piece, and the switches representing the connections.

Mechanical Design:
The puzzle was made out of ¼" thick black acrylic. We lasercut pieces out of the acrylic and made hollow boxes as each puzzle pieces. The boxes were hollow so they could hold the electronics. Over the course of manufacturing, we learned about the tolerance on the laserCAMM being greater than our application allowed for. Not accounting for the tolerance on the laser cutter made assembly a lot more difficult for us than it would otherwise have been. The inaccuracy in our build also caused our electrical contacts between pieces to be less stable than optimal.
For the contacts themselves, we used dome contacts from Snaptron, Inc. that compressed to a flat connector when pressed with adequate force to form a stable contact. We also made fake connectors on each puzzle so as not to give information on the puzzle through the connectors. With our tolerance issues, the contacts would not always form a stable connection giving us intermittent unintended vibrotactile feedback as we put the puzzle together.

Educational Demonstration

During the demonstration, many people tried out the puzzle in groups of three to six. With the haptic feedback, a most users were able to put together the puzzle pieces. We could not find specific findings in terms of the educational effect of the haptic feedback since our demonstration was not a controlled experiment.

Results

We were able to build "THE CUBE" after a few iterations on our electrical design. The electrical design worked consistently whenever the contacts were made. On our mechanical design, we were not able to get as much fake connectors on the puzzle so as not to give information on the puzzle away using the puzzles. The electrical connections between pieces were not as stable as we'd hoped for which caused us to have intermittent unintended vibrations.
Users at the Open House told us how much fun they had playing with the puzzle, and awarded our project with the vote for "Most Creative".

Future Work

Future work on this project would involve both manufacturing improvements and spatial intelligence testing with our target age group. If we were to rebuild this particular puzzle, we would make improvements such as:
1) using cross-hatching segments to interlock the pieces instead of just glue,
2) creating a printed circuit board for faster and easier manufacturing of the identical motor circuits for each piece,
3) use thinner acrylic (or masonite) so that the motor vibrations are more easily felt.
Next, we could build puzzles of different designs or create modular mini-cubes to be rearranged in any fashion.

The most important educational phase of our future work would include testing if and how this positive vibration feedback is helpful for children to pick up spatial intelligence. One method of measuring this is to give a diverse group of children (some with learning disorders) identical pre- and post-test of spatial orientation, on paper. In between the two tests, they would have significant play-time with THE CUBE, or a similar vibration feedback puzzle. This could involve free play or a more structured activity with specific challenges described in Jacoby. We would then see if there is a marked improvement on written test performance.

Acknowledgements

We would like to thank Snaptron, Inc. for their support with sending us the dome connectors we used for "THE CUBE". We would also like to thank Prof. Okamura, Nick Colonnese, and Ann Majewicz for the opportunity to work on the project.

Files

Bill of Materials
CAD Models and Drawings

The CAD files of most likely interest will be the ones with the _LaserCAMM ending, as these are the final piece designs for each of the six puzzle pieces. Our laser cut drawings with planar pieces are called "Cut X", where X is the sheet number. These are SolidWorks 2013 version files.

References

1) S. Jacoby, N. Josman, D. Jacoby, M. Koike, Y. Itoh, N. Kawai, Y. Kitamura, E. Sharlin and P.L. Weiss. Tangible user interfaces: tools to examine, assess and treatdynamic constructional processes in children with developmental coordination disorders. Proceedings of the 6th Internationall Conference on Disability, Virtual Reality & Associative Technology, Esbjerg, Denmark, 2006. http://www-human.ist.osaka-u.ac.jp/ActiveCube/AC_children_assessment.pdf

2) Bernhard Weber, Simon Schätzle, Thomas Hulin, Carsten Preusche and Barbara Deml. Evaluation of a Vibrotactile Feedback Device for Spatial Guidance. WHC 2011. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5945511

3) Maria Lehnung, Bernd Leplow, Lars Friege, Arne Herzog, Roman Ferstl, Maximilian Mehdorn. Development of spatial memory and spatial orientation in preschoolers and primary school children. British Journal of PsychologyVolume 89, Issue 3, pages 463–480, August 1998. http://onlinelibrary.wiley.com/doi/10.1111/j.2044-8295.1998.tb02697.x/abstract

4) Felix Ritter, Bettina Berendt, Berit Fischer, Robert Richter, and Bernhard Preim. Virtual 3D Jigsaw Puzzles: Studying the Effect of Exploring Spatial Relations with Implicit Guidance. http://www.eecs.tufts.edu/~mpoor01/DiscertationStuff/Assembly%20Task/Virtual%203D%20Jigsaw%20Puzzle.pdf

5) Tamara M. Lackner, Kelly Dobson, Roy Rodenstein, Luke Weisman. Sensory Puzzles, CHI '99 Extended Abstracts on Human Factors in Computing Systems. http://dl.acm.org/citation.cfm?id=632716.632882

6) Harders, M., Barlit, A., Akahane, K., Sato, M., Szekely, G. Comparing 6DOF haptic interfaces for application in 3D assembly tasks. In: EuroHaptics. (2006) 523–526. ftp://oenone-v06.ee.ethz.ch/publications/proceedings/eth_biwi_00438.pdf

Appendix: Checkpoints

Checkpoint 1

Our original goals for this checkpoint were to:

We can procure wiring, eccentric rotating mass motors, and other small electronic components from the SPDL. We can buy acrylic from the PRL and Tap Plastics, and cut it at the PRL or TLTL. Any other components such as connectors, we were able to find at RadioShack.

1. a) Mechanical Design

We began with a concept of a puzzle of the Golden Gate bridge, but this turned out to be impractical because there were so many thin lateral members-- not ideal for a 3D puzzle.

Our electronic plan of layering pieces lent itself better to a simple pyramid design, but this had many angles that were difficult to reproduce with planar, LaserCAMMED pieces.

This led to a more square design, a castle.

We ended up modeling our puzzle after a six-piece cubic eraser per Professor Okamura's suggestion. This was ideal because it was a difficult puzzle, but only had six pieces and two layers beyond the base, which was easier to build.

We created a scaled-up CAD of this eraser along with a base to hold the arduino and power source. Each side of the cube in our real model will be eight inches.

Each piece will be built out of planar, 1/4" acrylic pieces as shown below. We will use acrylic glue to hold the pieces together.

As shown, the pieces do not yet have holes for electrical connectors, because we are still experimenting with different types of connectors. If we use 9V battery connectors, these holes will be square, although these are not ideal because they are difficult to undo. Next, we will try dome switches, which will require circular holes.

1. b) Electrical Design

Original design schematic:

The first electronics design we conceived was designed to use event driven programming to turn a MOSFET on or off to allow current through a motor whenever a new piece was connected. When a piece is connected, an input to the Arduino is pulled up to 5V and an output is correspondingly set to 5V for a short period of time. This design called for a pyramid configuration of connectors, with the piece farthest away from the base (piece 2 in this diagram) needing four connectors and every piece below it needing two additional connectors. The row of switches in the schematic represents a boundary between pieces.

Revised Schematic:

The next iteration of our electronics design was meant to minimize the number of connectors needed to cause a short buzz at every correct connection. It used a high pass filter to allow for the change in a voltage level to be seen on the motor but did not allow the DC aspect of the voltage through. This approach only needed two connectors on each level.

2. and 3. Prototyping and Initial Testing

Here is a prototype of the wiring between the base and the piece:

We were able to implement vibration for a short period of time after the two "pieces" were connected. This was possible because we had one output signal for the arduino commanding the motor and one input signal identifying a connection. When a connection was made, a timer started and the vibration ran for a few seconds. The arduino code for a single motor, which can be easily implemented for 6 using 12 of the arduino pins, is below:

int motor = 8; // pin for commanding the motor
int signalPin = 3; // pin to determine if connection is made b/t pieces
int previousSignal = 0; // boolean for connection made in previous loop (high or low/0)
int timer; // time in ms since connection has been made
double signal; // boolean for connection made in the current loop (high or low/0)

void setup() {

  pinMode(motor, OUTPUT);
  pinMode(signalPin, INPUT);

}

void loop() {

  signal = digitalRead(signalPin);
  if (signal == HIGH && previousSignal == 0) {
    timer = 0;   // connection has just been made, timer begins
  }
  if (timer < 4000) {
    digitalWrite(motor, HIGH);  // connection made recently, motor turned on
  }
  else {
    digitalWrite(motor, LOW);  // motor off in all other cases
  }
  timer++;
  if (timer > 5000) {   // if timer value gets too high, data overflow
    timer = 5000;
  }
  previousSignal = signal;

}

Checkpoint 2

Our goals for this checkpoint were to:

1. Laser cutting

We added to our CAD models holes for electronics and motor mounts. We also came up with a design for a motor mount to fit the ERMs we were given, which is shown in the middle of the first sheet below on the left-hand side. These are all the pieces that make up our puzzle, which we sent to the LaserCAMM:

These are some of our cut pieces:

2. Code expansion

Since we have redesigned our electronics such that the motors are to be simply triggered by circuit completion alone, without a controller, this checkpoint goal no longer applies as we will not need to code motor timing into our arduino.

Other progress

Snaptron Electronics has kindly donated 100 FD14700 dome connectors, which will work well for the connections between puzzle pieces since they are easily mountable (we simply place them in the pre-cut holes), and they require compression to work (which will make them robust). Thank you Snaptron!

Dimensions shown: A = 0.551", B = 0.113", C = 0.037", D = not specified, E = 0.225", X = 0.5"

We have many more of these connectors than we will actually use so that the users cannot discern puzzle piece placement from connector location alone-- they will have too many fake ones to tell the difference.

Checkpoint 3

Our goals for this checkpoint are to:

1. Device construction and testing

We have taken the following steps in our construction:
1) glued each puzzle piece and the base together,
2) attached all the dome connectors that will carry power to the vibration motors in the circuits,
3) soldered the circuits onto electronic prototype boards and mounted them inside the pieces using small, cut acrylic pieces. We also mounted eccentric mass motors by simply gluing them to the inside walls of the puzzle pieces. The small motors from the Neutrino kit were not nearly strong enough to penetrate the 1/4" acrylic, so we used larger, more powerful ones from the SPDL.

Because the laser does not cut exactly to tolerance, we have had trouble getting one of the puzzle pieces to fit into the cube. Here are some pictures of the assembled puzzle so far:

Each piece works in that it vibrates when it is stacked correctly, but as of now the puzzle must be held in place or it slides around on the base and loses the connections easily. We have taken care to avoid false positives in vibration feedback by using a unique set of orientations of 5V and ground connections for each point of electrical contact with each side piece and the bottom and top pieces. We will also define the first piece to be placed so that only one puzzle solution is possible.

We have a few things to accomplish before presentation time:
1) get the last piece to fit and one other one to fit more easily. We will first look for Loctite glue dissolver so we can shift small parts of the puzzle pieces. If this does not work, we will have to hand sand the pieces down. This damages the finish and is not preferable.
2) clean up the glue residue off the pieces. We can use the glue dissolver and the buffer in MERL.
3) create a safer connection to the power supply (the same one as we used for the haptic paddles). Right now there are two wires sticking out of the base to which we alligator clip connections to the power supply.
4) glue down acrylic pieces on three of the corners where the first piece will be placed on the base. This will prevent sliding and create robust connections.
5) glue filler acrylic pieces into holes previously intended for motor mounts, which we no longer need.
6) glue the top of the base down.

Exploring other designs will likely be left as a "future work" item for anyone interested in improving upon our project.