The HumanoidLabInvolving Students in a Research Centre Through an Educational Initiative

Guillem Alenya1, Jose Luis Rivero2, Aleix Rull1, Patrick Grosch1 and Sergi Hernandez1

1Institut de Robotica i Informatica Industrial (CSIC-UPC), Llorens i Artigas 4-6, 08028 Barcelona, Spain2Open Source Robotics Foundation, 419 N Shoreline Blvd, Mountain View, CA 94043, U.S.A.fgalenya, arull, pgrosch, shernandg@iri.upc.edu, jose.luis.rivero.partida@gmail.com

Keywords: Project based Learning, Mentoring, Humanoid Robots, Introduction to Research.

Abstract: The HumanoidLab is a more than 5 year old activity aimed to use educational robots to approach studentsto our Research Centre. Different commercial educative humanoid platforms have been used to introducestudents to different aspects of robotics using projects and offering guidance and assistance. About 40 studentshave performed small mechanics, electronics or programming projects that are used to improve the robots byadding features. Robotics competitions are used as a motivation tool. A two weeks course was started thathas received 80 undergraduate students, and more than 100 secondary school students in a short version. Theexperience has been very positive for students and for the institution: some of these students have performedtheir scholar projects and research in robotics and continue enrolled in the robotics field, and some of them arecurrently in research groups at IRI.

1 INTRODUCTION

The Institut de Robotica i Informatica Industrial(IRI) is a joint research centre between Spanish Re-search Council (CSIC) and Universitat Politecnica deCatalunya (UPC). In 2007, a SWOT analysis revealedthat there was a lack of capacity to involve new stu-dents, principally undergraduate and Master students.IRI is located into one of the UPC university campus,but it was relatively unconnected from university ac-tivities and the visibility of IRI research areas in theuniversity was very small.

With this in mind, in 2007 the HumanoidLab ini-tiative was created with these objectives:

� Introduce engineering, compute science, andmathematics students to robotics.

� Promote multidisciplinarity in students’ educa-tion.

� Increase the number of students in the IRIRobotics Laboratory.

� Increase the visibility of IRI research activities inthe university community.

� Contribute to the community with open softwareand hardware projects.

Undergraduate educational robotics can be di-vided in three areas: robotic clubs, formal university

courses, and collaboration into a research project. TheHumanoidLab initiative combines a bit all three.

In the first place, some Universities and Researchinstitutions host robotic clubs grouped under dif-ferent initiatives, like student sections of the IEEEAerospace and Electronic Systems Society (AESS) 1

or the Board of European Students of Technology(BEST) 2. These clubs are maintained by students andusually base their activity in organize or participateat national or international robotic competitions, likeEurobot 3 (that takes place from 1998). This com-petitions propose a set of challenges, and each teamdesigns and builds custom robots with different me-chanic, electronic and software solutions. The Hu-manoidLab wanted to use competitions as motivationtool, but not as primary objective.

In the second place, regular robotic courses in-volving engineering topics are common (Matari,2004; Rus, 2006), where simple robots are assembledand programmed with basic reactive algorithms. Agood idea to focus the efforts in software developmentis to use standard and non-expensive robot platforms.In this line, undergraduate experiences using LEGOand AIBO have been reported (Sklar et al., 2007),

1http://ieee-aess.org/2http://www.best.eu.org/3http://www.eurobot.org/

213

and more recently using Roomba based robots likethe TurtleBot (Gerkey and Conley, 2011). However,in these platforms is sometimes difficult to embedmore sophisticated computer science topics that un-derlie perception, planning, and control mechanismsin modern robots (Touretzky, 2012). We liked the ideaof using robot kits because they provide a quick startplatform for people interested principally in program-ming, but we also wanted to keep the freedom to cus-tomize and replace mechanical and electronic parts,and embedded software.

In the third place, there are lots of students col-laborating on scientific projects at university depart-ments, but this is usually not performed as a formaleducational activity further than master theses andundergraduate works. Besides, departments and re-search groups have the capacity to participate in com-petitions that require complex and expensive equip-ment. Maxwell (Maxwell and Meeden, 2000) re-ported an intensive summer project where 10 un-dergraduate students should collaborate to develop acomplex robot used to serve hors d’oeuvres in a so-phisticated manner. The RoboCup competitions 4,founded in 1997, promote teams involving studentsnot only on the classical soccer competition but morerecently the rescue leagues (from 2001) and in therobocup@home (from 2006). On 2010 it engagedmore that 3000 participants from 500 different teams.We considered these activities interesting becausethey are long term projects where hardware and soft-ware platforms are evolved during different years andwhere collaboration and code reuse is promoted. Wealso wanted that the students could have the oppor-tunity of develop their Master thesis, undergradu-ate projects, and collaboration grants under the Hu-manoidLab.

In this paper, the HumanoidLab initiative is de-scribed and the results of 5 years of activity are eval-uated.

2 METHODOLOGY

The HumanoidLab group decided to use small hu-manoid robots in all the activities. Time has revealedthat this was a key decision because provided us witha distinctive position. In our area of influence, therewas a moderate activity in university and bachelor de-grees using wheeled robots, but we were the first us-ing humanoids. Humanoids were (and are) a growingtopic in the robotics community and give the oppor-tunity to work both in software and hardware and do

4http://www.robocup.org/

(a) Original Robonova robot (b) Twiki robot

Figure 1: Twiki robot includes many modifications devel-oped as student projects, as additional d.o.f. at legs, pan-tiltunit, foot pressure sensors, color camera, Gumstix process-ing unit, improved battery system, and a new C program-ming library.

it in a wide range of different areas: mechanics, elec-tronics, perception, locomotion, kinematics, manipu-lation, etc. International market offers the possibilityto get a fully working educational small humanoid bya reasonable price (800Eur) and consequently with areduced funding the HumanoidLab could start theiractivities. Most of the activities use these educationalhumanoid robots as a base, and foster their capabili-ties by adding new features.

The project has 4 project administrators, each onespecialized in a different area: mechanics, electronics,low level programming, and perception and planning.Each administrator is the responsible for an activityand for different groups of students. As robotics ismultidisciplinar, the tight collaboration between dif-ferent administrators is usually required and is ac-tively promoted in the group. The HumanoidLab isdeveloping 5 different kinds of activities that are de-tailed below.

2.1 Small Projects

This is the main activity. It involves a small group ofstudents, between 2 and 3, and it is performed dur-ing the scholar University period. We organize meet-ings between students and some of the administratorsto determine the previous knowledge of the students,the topic that they would like to work with, and thetime that they could spent. Project administrators pro-pose small projects that fit in the current Humanoid-Lab interests, and students can also propose their ownprojects. Once an agreement about the project scopeand the execution time has been reached, the projectstarts with the supervision of one administrator. Eachnew project provides new capabilities to incremen-tally obtain more complex and capable robots. In sec-tion 3 a detailed list of some of the carried out projectsis presented.

CSEDU�2014�-�6th�International�Conference�on�Computer�Supported�Education

214

2.2 Competitions

Competitions are not a goal by itself, but an oppor-tunity to put in practice and show the progress ofsome projects. The HumanoidLab regularly presentssome student teams that after the competition writea report (Pons et al., 2010). As it is not a cen-tral goal, can happen that the proposed solutionsare not the best suited to win the competition, butserve to demonstrate new developments. A clear ex-ample is the Twiki robot, who participated at the2009 CEABOT competition. In the stairs challenge,where stairs have to be recognized and traversed upand down, Twiki used embedded computer visionfor obstacle avoidance and pressure sensors to detectdownstairs (Pegueroles and Simo-Serra, 2010) (seeFig. 1(b) and 2(a)), while other competitors used sim-ple and effective infra-red sensors. However, it wasrecognized there as the most technological evolvedrobot and get the second prize.

In the Spanish national scope, small humanoidscompetitions are mostly reduced to one or two tour-naments per year with events distributed on severaldays. Each small humanoids contest is composed byseveral challenges, which can be group into two maincategories, depending on what is especially requiredto the robot:

� Mechanical skills: events designed to probe therobot mechanic design and actuators. An exampleof this kind its the stairs competition or the sumocontest.

� Perception abilities: events oriented to resolveproblems using mainly the robot sensor capacityon the environment. Obstacle race is a good ex-ample of this category.

2.3 Introduction Course

UPC University offers the possibility to include aworkshop or activity as part of its studies plan. Get-ting directly into the University studies and partici-pate as an active member has been a good opportu-nity to have contact with students. A two weeks in-troductory course/workshop was designed, presentedand finally approved by the University. That was thebirth of ”Introduction to Humanoid Robotics” course.Practice paid a role as important as theory. Partic-ipation of students from different schools was oneof the strongest proposal points so multidisciplinaryteams could be composed. The kind of robots used toperform the practical assignments has evolved fromthe beginning, and currently from 8 to 10 customizedBioloid robots are used. A maximum of 2 studentsfor each available robot that we have is allowed to

(a) (b)

Figure 2: Pressure sensors and pan-and-tilt unit on aRobonova robot.

join this course at each edition. The course takes4 hours by day, that is a total amount of 40 hoursper course. Past students came from Computer Sci-ence, Mechanical Engineering, Electronics, Mathe-matics and Physics areas.

After the great success of this activity we havebeen asked to propose a shorter introduction activ-ity, of about 2 hours of duration. We have used thisshorter activity as outreach in local conferences, sec-ondary school visits, and the European robotics weekactivities. We have made an effort to develop librariesand tools to facilitate the access to this technology toall kinds of students. Nowadays, the requirements toparticipate in this course are very basic: knowing theconcepts of variable, conditional and loop.

Based on the Introduction course we have also for-malized a course in a set of 5 sessions for secondaryschool teachers, that gives them knowledge and pro-poses activities to introduce robotics in the secondaryschool curriculum.

2.4 Curricular Projects

After its participation in some of the Humanoid-Lab activities, some students decide to perform theircurricular projects with us, and sometimes in otherrobotic groups at the IRI institute. They are basicallythe final career project (PFC) and the Master Thesis(MTh), but also some undergraduate courses assign-ments. Some of the works that where performed di-rectly in our group are detailed in Sec. 3.

2.5 Outreach Activities

We have participated in some outreach activities. Wehave performed some talks at Open Software confer-ences and to secondary school teachers. we have par-ticipated at robot exhibitions like Robot exhibition atSonar 2010 (Barcelona), IRC 2010 (Korea), and Fi-comic (2012). We have appeared on TV and radioprograms, and also appeared several times in localnewspapers. We regularly participate in the ScienceWeek and in 2012 also in the EU robotics week.

The�HumanoidLab�-�Involving�Students�in�a�Research�Centre�Through�an�Educational�Initiative

215

(a) IR foot sensors todetect up and downstairs

(b) Rotating IR sensor at thehead

Figure 3: Two different modifications on the Bioloid robot.

Figure 4: New sensor board for the Bioloid robot.

3 PROPOSED PROJECTS

In five years there have been more that 20 smallprojects performed. These involve different skills andstudents have the opportunity to learn different as-pects of robotics. A list of some them with codeis available on the developers site of the project 5.Here we list some of the most representative, from thebeginning of the project to the latest developments.Along each project it is specified whether it was anstudents project, a grant, curricular works or coordi-nators project. We have identified that keeping coor-dinators motivation is important, and maintain somechallenging projects is a good mechanism for such.

3.1 Robonova

The Robonova platform is a 16 dof robot (Fig. 1(a)).It has an embedded micro-controller and can be pro-grammed using a basic-like programming language.It has some analog channels available to use simplesensors.

Inverse Kinematics (students project). This wasone of the first activities, defined as a software project.The objective was to define the equations of the

5http://apollo.upc.es/humanoide/

Inverse Kinematics (IK) of both arms and legs ofthe robot, and to implement the corresponding algo-rithms.Motivation. IK is important to switch from motionsbased on pre-computed positions to the complete con-trol over the motion range.Organization. A group of three students was formed,with complementary knowledge on mathematics anengineering. The project required about 150 hours ofwork, and 20 hours of supervision based on regularmeetings.Evaluation. Administrators did not correctly evalu-ated the previous knowledge of the students, and theproject required more time than expected. Studentwere moderately satisfied because they had the op-portunity to learn the mathematical background of se-rial robot kinematics, and the different solutions. Onestudent left the group, and a new project was assignedto the other two in a pure software project followingtheir concerns. An issue was that the project could notbe fully completed because the algorithm could notrun on the embedded micro-controller. Consequently,the final evaluation was performed in simulation. Thismotivated the definition of a new electronics projectfor the addition of a new computing unit.

Extra Dof for Robot Legs (student project). Thisis a hardware project to modify the leg architectureto allow the inclusion of an extra rotational degree-of-freedom while maintaining almost the same robotheight (Fig. 1(b)). This was necessary to not compro-mise the overall equilibrium.Motivation. The original Robonova robot has 5 dofat each leg (Fig. 1(a)). Consequently, turns are onlypossible taking advantage of foot slippage, that is im-precise and with low repeatability.Organization. The project was assigned to one stu-dent with interest in mechanical engineering. Theminimum requirements were defined, and the studentpresented different alternatives that were evaluated,and the best one was implemented on a real robot.The project required about 200 hours of work, and 20hours of supervision based on regular meetings. Thehelp of the workshop personnel (20 hours) was cru-cial.Evaluation. The evaluation was based on the ob-tained mobility executed by the real robot, and the re-peatability of the turning motion. The student foundthis project as a great opportunity to improve his abil-ities in mechanical design an related software. Hecontinued in the group, and is now one of the super-visors. The design was published as open source inour website, and several other groups presented simi-lar designs in the next robotics competition.

CSEDU�2014�-�6th�International�Conference�on�Computer�Supported�Education

216

Feet Pressure Sensors (student project). In a previ-ous project a new foot to place force-resistance sen-sors (FSR) in the bottom and the front, and its as-sociated electronics (Fig 2(a)) was designed. In thisproject, and thanks to a CSIC grant, a student per-formed a formal study of the most convenient positionand and physical mounting of the sensors, and devel-oped a calibration hardware and software (Barbadilloand Alenya, 2010).Motivation. Pressure sensors were used satisfacto-rily in the skill of climbing up and down stairs. How-ever, we wanted to used them to develop walking al-gorithms, and more precision was required.Organization. A single student performed theproject. The project was defined taking into accountthe student background in sensors and electronics. Itrequired 300 hours of dedication. The supervisionwas based on regular meetings and took 10 hours.Evaluation. A new evaluation method for us was in-troduced in this project, based on defining 4 differentintermediate checking points with the student. Themethod worked very well, and let to assess differentevolution stages and evaluate the student . It is worthto mention that the knowledge acquired by the stu-dent was used in a subsequent research period he per-formed in another institution (Barbadillo et al., 2011).

3.2 Bioloid

The Bioloid robot is a 18 dof humanoid robot (seeFig. 3) made of modular parts that can be easilychanged. Compared to the Robonova, the motorshave more torque, and are protected against overloaddamage. The micro-controller has also basic compu-tation capabilities, and has some analog channels forconnecting simple sensors. It come equipped with asmall inertial unit intended to aid to the walking mo-tions. The robot is relatively cheap (� 800$) so wehave 12 of such robots that are used in small projectsbut also in the Introduction to Humanoids course andrelated outreaching activities. A project was proposedto add a new extension board (see Fig. 4) to expandthe robot controller with different buses and mountnew sensors, like a compass. Currently this projecthas been transferred to a local company that will com-mercialize it.

Human Motion Transfer (Master thesis). Thisproject uses a depth camera, based on the Time-of-Flight principle, to acquire human motion of an op-erator and transfer that motions to a Bioloid robot.The student had to implement vision algorithms, 3Dsegmentation, and implement a communication chan-nel between the vision computer and the robot. Later,

with the appearance of Kinect based sensors, the hu-man acquisition part has become more easy, and oth-ers have reproduced this work using this sensor (Pfeif-fer, 2011).Motivation. The main objective is to develop a mech-anism to create new robot movements that are human-like and easy to create.Organization. The project was assigned to a studentwith interest to learn image processing algorithms andinterest in control. The project required 400 hoursof dedication divided in 3 different stages: 150 hoursworking in the perception and body segmentation, 50hours for the adaptation of the IK algorithms and thecommunication with the robot, 150 hours developingthe graphical interface to acquire and reproduce mo-tions, and the rest for writing. It required 35 hoursof supervision, and 10 extra hours to solve problemswith the image segmentation and robot communica-tion.Evaluation. This was a Master Thesis project. Thestudent published a report (Siscart, 2011), and thework was publicly presented and evaluated by a jury.The student received the highest qualification. Thestudent was very satisfied because the project entailedclosing the perception-decision-control loop, and ithad the opportunity to learn perception algorithms.

3.3 Darwin

The Darwin robot (Fig 5) is a 18 dof platform that in-cludes one micro-controller, an embedded PC runningLinux, a color camera and an inertial unit. The hard-ware and the software is open source, and the inter-action with the robot is using a C library. Comparedto Bioloid, Darwin has a complete PC inside and itis capable of using more complex sensors like colorcameras. The architecture is also more robust. How-ever, it costs approximately 10 times the price of a Bi-oloid. We use this platform for small projects, but itis not possible to have several and use them in the In-troduction course. Again, with this platform it is alsopossible to propose mechanical, electronic, and soft-ware projects. For example, in the sensor part a depthcamera has been added(Fig 5), and different projectshave proposed different solutions to include grippers(Fig 6(a)).

ROS Integration (coordinator project). Currentlythe ROS framework is widely used and many robotsare supported, among others small humanoids. Inthis project the necessary layers to interface the Dar-winOP robot with ROS have been developed.Motivation. The integration of the robot in the frame-work enables the use of state-of-the-art algorithms

The�HumanoidLab�-�Involving�Students�in�a�Research�Centre�Through�an�Educational�Initiative

217

implementations compatible with ROS, for examplefor walking (Hornung et al., 2012).Organization. This project was a coordinator initia-tive, and assigned to a coordinator because it was along-term strategical project. It required 200 hoursto obtain the first working version. In periodic co-ordination meetings the evolution was presented anddiscussed with the other coordinators. The softwarehas been published under LGPL license, followingthe philosophy of the HumanoidLab. This project isstill alive, and requires 5 hours/month of a coordinatorfor maintenance and to include new features.Evaluation. The software has been presented in aDarwin competition 6. Other projects using this soft-ware are being proposed, and are used to evaluate andoccasionally propose new improvements. The soft-ware is also being used by other groups and we re-ceive periodically questions and requirements for im-provements that we try to answer.

3.4 Other

Leg Redesign (Master thesis). In this project we in-vestigated how to build a custom pair of legs usingpowerful motors. The student had to design all theframes and mechanical parts, built the legs, and pro-grammed them to obtain examples of walking gaits.The overall size of the robot was 70cm.Motivation. Evaluate the complexity and cost ofcreating custom robot legs to build new robots fromscratch.Organization. The project was assigned to a studentwith interest in mechanics and hardware abilities. Theproject required 400 hours of dedication, 30 hours ofsupervision, and the assistance of the workshop per-sonnel to use the machinery and tools to build the pro-totype.Evaluation. This was a Master Thesis project. Thestudent published a report (Cortada, 2010), and thework was publicly presented and evaluated by a jury.He received a high qualification. The evaluation in-cluded the demonstration of walking gaits and maxi-mum weight capacity measures.

4 FUNDING

To support the activities of the HumanoidLab we havereceived different kinds of funding. This has been pri-marily used to acquire robotic platforms and equip-ment like computers, additional sensors, and parts,but also to travel to robotic competitions.

6http://www.icra2012.org/program/robotChallenge.php

Figure 5: The DarwinOP robot has been modified with grip-pers and a depth camera. The control system has beencompletely rewritten to take advantage of the embeddedPC/micro-controller architecture.

(a) Simple gripper and addi-tional dof

(b) 3-fingered hand model cur-rently being implemented

Figure 6: Different developments for giving the DarwinOProbot the ability to grasp objects.

Generally students in the lab are volunteers. How-ever, we have obtained funding for grants in three dif-ferent programs. The first was from the regional gov-ernment and was intended to foster open source com-munity. We obtained two student grants. The sec-ond was from the Spanish Council of Investigationin a program to introduce undergraduate students toresearch. The third was a program to generate educa-tional contents oriented to secondary school. In total6 students have obtained a grant.

We started our activities with a reduced numberof robots, and we performed some with them smallprojects and developments. However, this changedwhen we decided to start the Introduction course andthe related outreach activities. More robots were

CSEDU�2014�-�6th�International�Conference�on�Computer�Supported�Education

218

needed to being able to make these activities, as weplanned to use one robot for each two students. Weobtained the rest of the robots we needed by buyingsome new units, but also with a donation from a localretailer.

The HumanoidLab has been mainly funded bya program of the Polytechnic University of Catalo-nia devoted to strength department activities with20KEur. Currently, the IRI is funding the develop-ments on the DarwinOP platform of the Humanoid-Lab with an internal project of 15KEur.

5 EVALUATION

The Robotics Introduction Course has had a great suc-cess. 80 students have participated in this intensive 2weeks activity. After each course the opinions of thestudents are summarized using reviews. Such reviewsreveal that before the course the knowledge about IRIactivities was 1 in a scale from 1 (unknown) to 7 (to-tally known) and before was evaluated as 6. The sameapplies for the knowledge of the HumanoidLab activi-ties. Concerning the evaluation of the course, studentsdeclare that will recommend the course to other col-leagues between 6 and 7. Of such students, 5 haveperformed their PFC in the Humanoid Lab and 1 inanother group.

The reviews also ask about knowledge acquiredin the course. Students consider that the conceptslearned will be useful in their career between 5 and6. The reviews also include specific questions aboutthe concepts divided in 4 sections: (a) kinematics andactuators, (b) sensors, (c) vision, and (d) robot pro-gramming. Questions ask for the knowledge beforeand after the course, and also the satisfaction after thecourse. Students declare in (a), (b) and (d) a previ-ous knowledge of 1-2, a present knowledge of 5-6,and a satisfaction of 4-5. Surprisingly, regarding (c)previous knowledge is declared the same 1-2, presentknowledge is declared low (3-4), but satisfaction isdeclared high (5-6). We believe that this is becausethe robots we are using are quite simple, but computervision are perceived as very powerful tools. Somestudents have asked, in a free comment space, for aspecific activity about computer vision for robotics.

The Introduction course has been served as a ba-sis for creating a course for secondary school teach-ers, and also a short version is being currently usedfor outreach activities of the IRI. More than 100 sec-ondary school students have participated in this activ-ity, always in collaboration with their teachers. Stu-dents are in general very satisfied, and their teachershave always expressed that the activity has been very

positive and helpful.More than 40 students have been involved in small

projects, and contacted the HumanoidLab thanks toword-of-mouth marketing. This has reported a widelist of improvements to our platforms. Some of thesenew features have been successfully adopted by othergroups working with the same robotic platforms.

Women in engineering and robotics are a minority,at least in Spanish education system. Surprisingly, ofthe 9 active students nowadays in the HumanoidLab7 are women. We have been asked sometimes for thereasons by researchers in Sociology, but unfortunatelywe do not have yet the answer.

We have some special cases that is worth to men-tion. Three former students of the HumanoidLab haveperformed its PFC in one research group of the IRI,and currently have become PhD students. One ofthem is still involved with us and has become admin-istrator. Additionally, 3 students have performed theirMaster thesis in the HumanoidLab.

We have tracks of 3 former students performingresearch activities related to robotics in research in-stitutions in other countries.

Currently, we have a DarwinOP robot and 12 Bi-oloid based robots working. Other robots have be-come obsolete and no further development is per-formed on them.

Even if it not an objective, participation in com-petitions has been very successful. The first year ourteams obtained the second and the fourth place at theSpanish humanoid competition. The next year wewin the second place again, and in the two last com-petitions the Humanoid Lab team has obtained thefirst place. These two teams are actually formed bywomen.

We believe that, for being useful to the objectivesof the HumanoidLab initiative, the participation incompetitions should be conceived with the followingobjectives in mind:

� Multidisciplinary collaboration: autonomous con-tests requires always from both, hardware andsoftware, development. They represent a goodmotivation to encourage collaboration betweengroups of students performing different tasks inorder to get a robot working or improve them.

� Promote group spirit: since Humanoid Lab hostsseveral group of students whom compose differ-ent teams, some measures were taken to ensureall people feels like belonging to the same orga-nization. It was the tutor mission to explain newteams that all the previous work they were usingwas made by their co-workers at IRI, but rivalsduring the competition. It’s a golden rule that anyprize won by any of the IRI teams is spent on

The�HumanoidLab�-�Involving�Students�in�a�Research�Centre�Through�an�Educational�Initiative

219

improving robots and laboratory material. Thisavoid any problem or ambition with respect to themoney and helps a lot on spreading the work re-sults sharing. For competitions which imply dis-placement to other cities, the own travel and coex-istence helps a lot to create bonds among all stu-dents, no matter what team they are.

� Provide long and mid-term explicit goals: the res-olution of different competitions events can in-spire both, tutors and students, in the search ofgoals to reach. From an specific competitionchallenge teams can start working on differentprojects, from mechanical design to artificial in-telligence algorithms, in order to be part of tour-nament competitors.

6 CONCLUSIONS

The HumanoidLab initiative was funded to introducestudents to robotics and give visibility to the researchactivities of IRI in the student community. After 5years, nearly 100 undergraduate students have partic-ipated in one of the organized activities. Of that, 40have participated in long term activities. Several ofthem have continued to work in robotics either in IRIor elsewhere. It is worth to mention that 7 of the cur-rent 9 students are women.

Using humanoid robots has revealed to be a gooddecision, because of the distinction with other activi-ties but also because the inherent problems are chal-lenging. Historically different humanoid robots havebeen used and improved using open hardware andopen source paradigms. The set of performed activ-ities include small projects, curricular projects, com-petitions, courses and outreach activities.

The synergies between mechanical and electronicengineering, computer science, and mathematicianshas been very positive and have been explicitly pro-moted.

Finally, as it is a volunteer activity it is importantto keep coordinator’s motivation. One suitable waywe have found is to propose challenging projects alsoto the coordinators that allow to learn from other ar-eas or expand their knowledge in new areas. How-ever, it has to be taken into account that long termprojects, suitable to be assigned to a coordinator, canrequire sustained dedication over the time, like soft-ware maintenance.

REFERENCES

Barbadillo, G. and Alenya, G. (2010). Diseno de un piepara un robot humanoide. Technical Report IRI-TR-10-08, Institut de Robotica i Informatica Industrial,CSIC-UPC.

Barbadillo, G., Dautenhahn, K., and Wood, L. (2011). Us-ing fsr sensors to provide tactile skin to the humanoidrobot kaspar. UH Computer Science 511, Universityof Hertfordshire.

Cortada, S. C. (2010). Requirements and design of a hu-manoid robot for playing soccer. Master’s thesis, Uni-versitat Politecnica de Catalunya.

Gerkey, B. and Conley, K. (2011). Robot developer kits.IEEE Robotics and Automation Magazine, 18(3):16–16.

Hornung, A., Dornbush, A., Likhachev, M., and Bennewitz,M. (2012). Anytime search-based footstep planningwith suboptimality bounds. In Proc. of the IEEE-RASInternational Conference on Humanoid Robots, Os-aka, Japan.

Matari, M. J. (2004). Robotics education for all ages. InProceedings, AAAI Spring Symposium on Accessible,Hands-on AI and Robotics Education.

Maxwell, B. A. and Meeden, L. A. (2000). Integratingrobotics research with undergraduate education. IEEEIntelligent Systems, 15(6):22–27.

Pegueroles, J. and Simo-Serra, E. (2010). Quadern de tre-ball ceabot 2009 - twiki. Technical Report IRI-TR-10-13, Institut de Robotica i Informatica Industrial,CSIC-UPC.

Pfeiffer, S. (2011). Guiado gestual de un robot humanoidemediante un sensor ”kinect”. Master’s thesis, Univer-sitat Politecnica de Catalunya.

Pons, S., Tolos, N., and Troyano, S. (2010). Algorismesi modificacions hardware a la robot dorami per par-ticipar al campionat nacional de robotica humanoideceabot 2010. Technical Report IRI-TR-10-12, Institutde Robotica i Informatica Industrial, CSIC-UPC.

Rus, D. (2006). Teaching robotics everywhere. IEEERobotics and Automation Magazine.

Siscart, M. R. (2011). Algorithms and graphic inter-face design to control and teach a humanoid robotthrough human imitation. Master’s thesis, Universi-tat Politecnica de Catalunya.

Sklar, E., Parsons, S., and Azhar, M. Q. (2007). Roboticsacross the curriculum. In Proceedings of the AAAI.

Touretzky, D. S. (2012). Seven big ideas in robotics, andhow to teach them. In Proceedings of the 43rd ACMtechnical symposium on Computer Science Education,SIGCSE ’12, pages 39–44.

CSEDU�2014�-�6th�International�Conference�on�Computer�Supported�Education

220