AI in Children’s Play with LEGO Robots
Henrik Hautop Lund
- 发表年份
- 2002
- 引用次数
- 8
摘要
We have made a number of robot applications as children toys. The design principles behind these applications are based on different artificial intelligence techniques. One of our main principles is to go away from the traditional play scenario of two agents such as child and intelligent computer game to a physical reality of the second agent. Hence, the game consists of a child playing with a physical robot that can be manipulated to different performances. For instance, we constructed a physical model of the old, popular computer game Pacman, where the child is to navigate a semi-autonomous LEGO Mindstorms robot around in a labyrinth while avoiding two fully autonomous ghost In other cases, we extend the games to include three agents: child, robot, and computer game, and in some cases, we even have four agents to interplay: child, robot, computer, and intelligent room. Introduction Educational Argument In the LEGO Lab, we are interested in how to apply artificial intelligence techniques in children’s play with LEGO robots. We find it essential, that the children have a physical entity to play and interact with, in contrast with the traditional computer games with which children interact with a virtual reality. Our attachment to the use of physical entities stems from our educational experiences with the use of robots. We find that children can learn about the real world (including math, physics, engineering, real world systems) by working with robots when using a constructionism approach (Papert 1980), especially when using guided constructionism approach rather than the unguided constructionism approach (Lund 1999). In the guided constructionism approach, lectures/guidance and hands-on experience is combined in order to allow the students to learn about a subject. In general, the constructionism approach puts emphasis on the hands-on experience, and it is believed that children can obtain a tool to think Copyright ©1999, American Association for Artificial Intelligence (www.aaai.org). All rights reserved. about an artifact by playing with/constructing the artifact. Here, we believe that it is important that the artifact is physically represented. Our educational experience shows us that if the artifact is virtual, then at least in a number of cases, students tend to use abstraction about different characteristics of the artifact abstractions that have no validity in reality! (Lund 1999) -see also (Miglino, Lund, & Cardaci 1999). Therefore, we believe that some subjects need to be taught in a way that puts emphasis on the child/student interacting with a physical entity. When we are making toys for our children, we should be very aware of the positive or negative educational impact that these toys may have on our children. Therefore, we should also take it serious if research on students’ use of abstractions shows that the abstractions lead to unrealistic views of real world systems. This should tell us that even though abstractions (and hence virtual realities) might be good in numerous circumstances, they might have a negative effect in other circumstances. We are trying to make a number of toy applications that allow children in a better way to ground their knowledge about artifacts in the real world. Experiments LEGO Robot Games In the LEGO Lab, we have made a number of experiments that investigate the use of robots in children’s play. In the autumn of 1998, during the Danish National Science Festival, we had 190 school classes with approximately 4,000 pupils of the age 7-15 to play with different interactive LEGO Mindstorms robots developed at the LEGO Lab. The robots include both wheeled and legged LEGO robots among these an 8legged LEGO spider robot with a layer of autonomous behaviour (walking, obstacle avoidance, etc.) which functionality can be overwritten by the child when controlling the robot’s behaviour with a LEGO joystick. In this sense, the control system of the LEGO spider robot is a behavior-based system (
关键词
相关论文
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Artificial intelligence: a modern approach
1995
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991
A new optimizer using particle swarm theory
R.C. Eberhart, James Kennedy
2002