The Micro-Rato Contest: Mobile Robotics for All - CiteSeerX

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

The Micro-Rato Contest: Mobile Robotics for All Luís Almeida, Pedro Fonseca, José Luís Azevedo, Bernardo Cunha {lda,pf,jla,mbc}@det.ua.pt Dep. de Electrónica e Telecomunicações, Universidade de Aveiro 3810-193, Aveiro, Portugal Abstract The Micro-Rato Contest of the University of Aveiro is a competition among small autonomous and mobile robots. Organised by the Electronics and Telecommunications Department of that university (DETUA), the contest aims at promoting, in a festival-like environment, the practical and integrated use of several topics typically taught in electronics courses. The teams, mainly formed by undergraduate students, develop their robots aside the courses formal curricula. Along its past 5 editions, the Micro-Rato Contest has had a growing participation, either in number of teams as well as in audience. It is, today, a relatively well known initiative all over the country, particularly in universities and institutes offering degrees in robotics, electronics or computer science. This paper provides an overview of the contest, starting by a brief history covering its evolution. It also includes a small description of the rules, technical specifications and support, as well as a brief profile of some of the robots which performed best in the 1999 edition. Keywords : Mobile Robotics, Robot Competitions, Robotics in Education.

1.

M OTIVATION AND STRATEGY

The Micro-Rato Robotics Contest of the University of Aveiro has been organised by a group of people belonging to DETUA's teaching staff. Hence, the contest is naturally developed as a pedagogical activity that aims at complementing the technical education of the DETUA students in an informal and extra-curricular way. This is done at two levels. In one hand, building a Micro-Rato (the portuguese expression for micromouse) requires the integration of concepts taught in different subjects which are typically part of courses on electronics engineering and computer science. For example, it is normally required some level of knowledge on digital and analogue electronics, microprocessors (either programming and system integration), power electronics, instrumentation, digital control, signal processing, etc. On the other hand, voluntary team work is also promoted which is indeed a prerequisite to participate in the contest. The strategy that has been followed by the organising committee is essentially based on the use of competition as a way to improve the motivation and interest of the participating teams in building their own robots. Such competition is tempered by providing enough technical resources as well as by promoting group sessions for technical discussions. These group sessions contribute to establish a positive relationship both at the inter-team level as well as between the teams and the organising committee. Besides, the increasing media coverage of this simple festival-like event also plays an important role, both as a contribution to increase the teams motivation and as a visibility window which enables the Micro-Rato Robotics Contest as a tool to promote science and technology in the global society of today.

2.

BRIEF HISTORY

The Micro-Rato Robotics Contest inherited its name from the well-known Micro-Mouse Contests organised since 1984 by the IEEE Computer Society [1] and later also by other technical institutions. However, the technical options taken in this contest make it significantly different from the original one as explained further on. In fact, it is more like the contest organised in the scope of the MIT 6.270 course on Autonomous Mobile Robotics [2]. In the summer of 1995, the DETUA Masters Degree offered a course on Autonomous and Mobile Robotics taught by Prof. Keith Doty from the University of Florida, USA. The course, which included

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

several techniques to build small and inexpensive robots, was the triggering event that gave birth to the very first contest of this kind in the University of Aveiro and in the country. Some of the members of the DETUA teaching staff decided, then, to use such knowledge and resources to set up a contest which could contribute to complement the technical education of the DETUA students in general. The first contest took place in December 13th , 1995, and counted with 6 teams, one of which came from the University of Minho, Guimarães (UM). The next edition took place 15 months later, in March 12th , 1997. This time there were 14 teams registered from which 11 showed up at the contest. From these, two teams were also coming from the UM. The third edition was held in May 6th , 1998. The number of participating teams increased to 23 amongst 28 registered. The composition and origin of the teams also showed, for the first time, a great variety. There were teams of undergraduate students of various levels (one of which won the contest!), of PhD and MSc graduated students and of teaching assistants. Three teams came from the UM, one from the UC, Coimbra,

a) Participants in the inaugural edition (1995).

b) Participants in 1997.

c) Participants in 1998

d) Participants in 1999 Fig. 1 – Participating robots from 1995 to 1999.

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

Fig. 2 – View of the audience in the 1999 edition. another one from UNL, Lisbon, and finally two from IPCB, Castelo Branco. There were even two teams formed by DETUA ex-students. 50 45 40 35 30 25 20 15 10 5 0

Registered Participants

95

97

98

99

2000

Year

Fig. 3 - Evolution of the number of registered and participating teams. In May 19th 1999, the 4th edition of this contest took place. Among 41 registered teams, 28 showed up with amusing robots, some of which presented a reasonable level of complexity. Also, several demos of other autonomous and mobile robots were performed, turning the contest into a festival-like mobile robotics event. Figure 2 shows a vie w of the audience during the 1999 edition of the contest. Finally, the 2000 edition, held in May 17th, consolidated the national impact of this contest. 47 teams registered, from which 29 showed up. From these, more than a half came from outside Aveiro. 12 Teams where from other Universities, 6 from technical Schools, 2 were private, 1 came from a Secondary School and, for the first time there was a registered team from a spanish university, Universidad Europea de Madrid. Figure 3 shows the evolution of the number of registered as well as participating teams. Although there has been a steady growth since the first edition, it is thought that such growth will be reduced in the near future. The number of 47 registrations in the 2000 edition already seems to indicate that a more stable participation level is being reached.

3.

THE RULES

Although the contest rules have evolved since the first edition as a result of the gathered knowledge acquired through a four year experience, the fundamental aspects still remain the same. The robots compete in rounds of three in a closed area, 5x5m wide, called maze. The robots objective is to go from the starting point to the goal, as fast as possible, and without colliding with the obstacles in the way. Once in the goal area, the robot must stop and stay there until the round is over (i.e. when all the three robots have finished or the maximum time has elapsed). The floor is covered with infra-red (IR) reflective material, except for the goal area. The goal is a 90cm wide circular area, painted in black, in the centre of which stands a 30cm high omnidirectional IR beacon. The robots face two types of obstacles; they must avoid other robots, which (most of the times...) behave like moving obstacles, and avoid the boxes that are scattered all over the competition area. The boxes, as well as

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

Start

Goal

Fig. 4 – The maze used in the 1999 edition. the inner walls, are covered with IR reflective material (fig. 4 shows the maze used in the 1999 edition). Some of the boxes may be higher than 30cm (up to 50cm), therefore occluding the view of the beacon from some spots within the maze. Robot size is only limited in what concerns its maximum dimensions. The robots shall fit in a box 30cm wide (square) per 40cm high. This size limit is valid at any time. If the robot has moving parts, all possible configurations of the moving parts shall comply with this limit. Furthermore, an additional limit is imposed: no robot can be wider than 15cm (circle) from the height of 25cm upwards. This reduces the shade area, with regard to the beacon light, caused by the robots that park in the goal. The boxes used as obstacles are either touching each other or at least 40cm away from each other. Every penalty incurred by the robot is converted into penalty time. The ranking is established by adding this penalty time to the time counted from departure to arrival at the goal. There is a 3 minute maximal limit for each run (fig. 5 shows two of such runs). The competition is normally organised as a series of 4 rounds: the first two are taken by all robots, the second one is selective, and the final round is taken only by the 3 top robots to decide the winner.

a)

b) Fig. 5 – Robots competing in the 1999 edition.

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

Starting in the 2000 contest, the maze is no longer known to the contesters before competition starts. During each round the robots stay in a "closed park" and may not be changed. Between rounds, competitors have a short period of time to perform any technical adjustment they want after which the robots go back to the "closed park." Maze configuration can be altered for any new round.

4.

TECHNICAL SUPPORT / BIBLIOGRAPHY

A fundamental issue in this contest is the technical support provided to the teams. This support is implemented in two ways. On one hand, the teams have access to our component store, where they can get parts and components for the robots. On the other hand, technical sessions are regularly organised during the weeks preceding the contest, in order to help the teams in solving their ongoing problems. In what concerns parts for the robots, the Organising Committee can also provide, to those teams that request it, a basic kit of sensors to detect obstacles, a beacon and a goal area, a controller board with an IO expansion and the robot base, motors and wheels included. Software is also made available to the teams, namely libraries with functions for motor control, sensor reading and even a real-time kernel (ReTMiK - Real-Time Micro-mouse Kernel), which facilitates the management of concurrent and preemptive cyclic tasks. This is particularly useful when using behaviourbased programming techniques. Since 1997 the Organising Committee offers a technical book to each participating team. These books are always related to the construction of small robots and they are a good source for anyone wishing to build its own Micro-Rato (e.g. [3], [4] and [5])

5.

BRIEF OVERVIEW OF ROBOTS THAT PARTICIPATED IN THE 1999 EDITION

The 1999 edition was the first one that counted with two robots that used the notion of localisation in order to identify their present position in the maze while moving towards the goal. One of them used relative spatial information based on odometers coupled to the wheels. This robot (Spark), the winner of this contest, was able to accomplish a set of missions combining a reactive behaviour (e.g. avoiding obstacles and other robots) with a sequence a pre-programmed short and simple trajectories. The other robot (D. Dinis) used absolute spatial information based on the combination of an electronic compass and beacon readings (direction and intensity). By using a maze map, the robot could follow a pre-programmed path, leave it whenever necessary to avoid obstacles and return to it to reach the goal. This robot ranked second and also received a special prize for innovation. The remaining robots did not use any aprioristic information. Some of them (e.g. Bulldozer II, Dyno and Insónia ) used a strategy based on the concept of behaviour-based autonomous agents according to which the global conduct resulted from a combination of several basic reactive and autonomous behaviours associated with robot basic functions. Typical behaviours are beacon following, collision avoidance, wall following, arrival detection and parking. Other robots used a simple non-terminating loop where all the sensory, control and actuation functions were executed in sequence (e.g. Tripé II and Jerry). In terms of construction, several differences among the robots could be identified, mainly concerning the beacon detection system and the underlying micro-processor platform. Concerning the beacon detection mechanism, there were solutions based on sets of more than 4 fixed sensors, on one pair of fixed sensors and even on a single sensor, either fixed or rotating. In terms of the processing hardware the most common solutions were based on the 80C188 (Spark, Bulldozer II and Tripé II) for which the real-time kernel ReTMiK was available, on the 68HC11 (D. Dinis, Insónia and Jerry) and on the 80C51 (Dyno, in this case the processor was assisted by one FPGA). The robots mentioned above are briefly described in [6].

6.

CONCLUSION

In this paper the Micro-Rato Contest has been described. Although inspired on the earlier IEEE MicroMouse contests it uses different technical specifications that facilitate the robots construction. This fact combined with a significant support both for components and technical issues, as well as with the natural appeal of autonomous mobile robotics, led to a large participation which has been consistently growing over the years.

Presented at CONTROLO 2000, The Portuguese Control Conference, Guimarães, Outubro de 2000

Although based mainly on undergraduate students the participating teams also gather people with very diverse educational levels, from first-year undergraduates to graduate students, either MSc and PhD, including team members of the teaching staff as well as of ex-students now working in companies or other institutions. The 2000 edition also saw, for the first time, a secondary school team coming for participation. Furthermore, the option for relaxed, although not trivial, technical specifications, has allowed for the coexistence of a significant number of solutions, which span from very simple to rather complex ones. All these factors together have fostered the pedagogical effect of the contest beyond the mere isolated construction of a small mobile robot. Participants, while exercising their technical and team work skills, learn from other teams’ experience and get eager to learn more, and even try new, possibly more complex, solutions. Positive impacts of the contest on the students have been noticed on several areas from which the following are highlighted: the development of embedded applications, either SW, HW and noise issues, the integration of HW components, mainly sensors and actuators with microprocessor-based systems, the programming skills of basic reactive behaviours such as obstacle avoidance and beacon or wall-following, the perception of real-time constraints and the capability to use real-time tools. Hence, the Micro-Rato Contest is indeed revealing itself as a valuable tool to aid teaching in electronics and computer science. On the other hand, the nation wide impact of the contest and the promising interest from secondary schools let us foresee the consolidation of the event, in the coming years, as an yearly gathering of robotics loving people supported on a mix of science, engineering skill, creativity and a lot of fun.

7.

ACKNOWLEDGEMENTS

The authors would like to acknowledge all the motivation and support received from a relatively vast number of people and institutions to whom they owe the successful organization of the Micro-Rato Contests. In particular we would like to thank all those that have participated in the contest jury, giving their personal contribution to the contest rigorousness, namely professors Keith Doty, Carlos Couto, Isabel Ribeiro, Tenreiro Machado, Urbano Nunes, Pedro Lima, José Barata, Francisco Vaz, Alexandre Mota, Fernando Ribeiro and Artur Pereira.

8. [1] [2] [3] [4] [5] [6]

REFERENCES C.M. Tetta, The evolution of the Micromouse Competition, IEEE Potentials, May 1986. Introduction to 6.270, MIT 6.270 course notes, 1992. Joseph Jones and Anita Flynn, Mobile Robots: Inspiration to Implementation, A.K. Peters, 1993. Gordon McComb, The Robot Builders Bonanza: 99 Inexpensive Robotics Projects, McGraw-Hill, 1995. J. Conrad and J. Mills, Stiquito: Advanced experiments with a simple and inexpensive robot, IEEE Press, 1997. L. Almeida, P. Fonseca and J.L. Azevedo (eds.), Secção Especial: Concurso Micro-Rato’99, Revista do DETUA 2(6), September 1999. (http://microrato.ua.pt/docs/artigos/publicacoes.html) .