Presentación sobre la evolución de los robots, desde el autómata clásico a los actuales sistemas "emotivos", comentando la tecnología base en cada caso
3. Conceptos generales
• Robot (1979): manipulador reprogramable y multifuncional
diseñado para llevar a cabo una tarea específica a través de una
serie de movimientos programados (Robot Institute of America).
4. Conceptos generales
• Robot (1990): Máquina controlada por ordenador y
programada para moverse, manipular objetos y realizar
trabajos a la vez que interacciona con su entorno.
5. 1. Breve historia
El mito de Pigmalion. “Sobre el teatro automático”
Heron (280 aC)
Grecia clásica
Falta necesidad y tecnología.
6. 1. Breve historia
Robot antropomórfico
(DaVinci, 1490)
Fuente del pavo
Real (Al Jazari, 1206)
Pato de Vaucanson
(Vaucanson , 1750)
7. 2. Breve historia
1921: Capek usa por primera vez el
término robot en R.U.R.
1926: Fritz Lang presenta por primera vez
robots mecánicos en Metropolis.
8. 1. Breve historia
1942: Isaac Asimov escribe Runaround y I,
Robot. Inicio de la robótica como ciencia.
9. 1. Breve historia
1946: Devol y
Englenberger dise ñ an
los Unimates, que
transportan maquinaria.
1946: Presper Eckert y
Maulchy dise ñ an el
Eniac.
13. 1. Breve historia
1974: Scheinman funda Vicarm.
1976: Las sondas Viking
incorporan brazos robóticos.
14. 1. Breve historia
1976: Se comienza a
desarrollar el Mars
Rover.
A ñ os 90: Aparecen el
COG, el IT, el Lunar
Rover, el Dante, el P3,
el Soujourner…
15. 2. Robots autónomos
• Un robot se considera autónomo e inteligente cuando:
– El sistema de navegación reside en la propia máquina y es
capaz de operar sin conexiones físicas a equipos externos.
– Es capaz de tomar decisiones por sí solo a partir de la
lectura de sus sensores
16. 2. Robots autónomos
• El sistema de control de un robot autónomo puede ser:
– Deliberados: SPA-primero perciben, luego planifican y,
finalmente, actúan.
– Reactivos: asocian patrón y acción. Comportamientos más
complejos resultan de las interacciones de los sencillos.
– Híbridos: combinan los dos anteriores
18. 3. Robots adaptables
•La adaptación se basa en el aprendizaje, o
capacidad de resolver una tarea sin necesidad de
programarla explícitamente.
•El aprendizaje facilita a los usuarios no expertos a
incorporar robots a la vida diaria (~plug&play)
•El objetivo final del aprendizaje es decirle al robot
QUE tiene que hacer y no COMO hacerlo.
19. 3. Robots adaptables
Redes neuronales: modelan
funciones no lineales a partir
de un conjunto de datos
Logica difusa: permite
modelar conocimiento de
forma parecida a la humana
Razonamiento basado en casos:
Permite aprender a partir de
Experiencias previas
21. 3. Robots adaptables
Aprendizaje supervisado
• El sistema consiste en “enseñarle” al robot qué haría un humano
en su lugar manejándolo con un joystick.
22. 3. Robots adaptables
Aprendizaje por experiencia
• El sistema consiste en “enseñarle” al robot qué haría un humano
en su lugar manejándolo con un joystick.
24. 4. Robots evolutivos
• Vida artificial: Sistemas artificiales que exhiben
propiedades similares a los seres vivos, a través de
modelos de simulación.
• Christopher Langton fue el primero en utilizar el
término a finales de los años 1980 ("Primera
Conferencia Internacional de la Síntesis y Simulación
de Sistemas Vivientes“, Los Alamos National
Laboratory)
• Se define “vida” como una manera de
autoreproducción, almacenamiento de información,
evolución, crecimiento y adaptación.
25. 4. Robots evolutivos
COREWARS
•Un papel se copia a sí mismo lo más rápidamente, así sacrifica
velocidad de ataque por perdurabilidad.
•Una piedra directamente bombardea direcciones de memoria
intentando matar rápidamente al mayor número de enemigos.
•Una tijera son los que emplean estrategias sofisticadas.
•Un vampire o pit-trapper roba procesos a sus oponentes
alterando su código máquina en el bombardeo para que
pierdan ciclos.
•Un imp es tan pequeño que su tamaño lo hace difícil de
neutralizar.
26. 4. Robots evolutivos
JUEGO DE LA VIDA
•El juego de la vida es el mejor ejemplo de un
autómata celular, diseñado por el matemático británico John
Horton Conway en 1970.
•Permite observar cómo patrones complejos pueden provenir
de la implementación de reglas muy sencillas.
•Cada célula tiene 8 células vecinas. Las células tienen dos
estados: "vivas" o "muertas“:
Una célula muerta con exactamente 3 células vecinas vivas
"nace" al turno siguiente.
Una célula viva con 2 o 3 células vecinas vivas sigue viva,
en otro caso muere o permanece muerta("soledad" o
"superpoblación")
Bloque Barco Parpadeador Sapo Planeador Nave ligera
27. 4. Robots evolutivos
“The key to create intelligent robots consists of letting them evolve,
self-organize, and adapt to their environment”
Idea (1984): Valentino Braitenberg: “Robots en la mesa”
Primer experimento (1994): EPFL, Swiss Federal Institute of
Technology in Lausanne y la Universidad de Sussex en Brighton
32. 5. Robots sociables
A efectos de convivir con los humanos, se busca que los robots
presenten un comportamiento empático. Para ello, se analiza el
comportamiento humano a partir de: expresión facial, postura
corporal, gestos, dirección de la mirada y voz.
33. 5. Robots sociables
Color de piel.
Movimiento.
Detección de ojos.
Estimación de distancia.
Aproximación.
Estimación de velocidad.
Detección de sonido
Detección de habla
Estimación de tiempos
n order to carry out evolutionary experiments without human intervention, at EPFL we developed the miniature mobile robot Khepera (6 cm of diameter for 70 grams) with eight simple light sensors distributed around its circular body (6 on one side and 2 on the other side) and two wheels (figure 1). Given its small size, the robot could be attached to a computer through a cable hanging from the ceiling and specially designed rotating contacts in order to continuously power the robot and let the computer keep a record of all its movements and neural circuit shapes during the evolutionary process, a sort of fossil record for later analysis. The computer generated an initial population of random artificial chromosomes composed of 0's and 1's that represented the properties of an artificial neural network. Each chromosome was then decoded, one at a time, into the corresponding neural network whose input neurons were attached to the sensors of the robots and the output unit activations were used to set the speeds of the wheels. The decoded neural circuit was tested on the robot for some minutes while the computer evaluated its performance (fitness). In these experiments, we wished to evolve the ability to move straight and avoid obstacles. Therefore, we instructed the computer to select for reproduction those individuals whose wheels moved on a similar direction (straight motion) and whose sensors had lower activation (far from obstacles). Once all the chromosomes of the population had been tested on the robot, the chromosomes of selected individuals were organized in pairs and parts of their genes were exchanged with small random errors in order to generate a number of offspring. These offspring formed a new generation that was again tested and reproduced several times. After 50 generations (corresponding to approximately two days of continuous operation), we found a robot capable of performing complete laps around the maze without ever hitting obstacles. The evolved circuit was rather simple, but still more complex than hand-designed circuits for similar behaviours because it exploited non-linear feedback connections among motor neurons in order to get away from some corners. Furthermore, the robot always moved in the direction corresponding to the higher number of sensors. Although the robot is perfectly circular and could move in both directions in the early generations, those individuals moving in the direction with less sensors tended to remain stuck in some corners because they could not perceive it properly and thus disappeared from the population. This represented a first case of adaptation of neural circuits to the body shape of the robot in a specific environment.
Two robots, a predator and a prey, are co-evolved within a square arena. The predator has a short-range sensors (1 cm) and a vision system, whereas the prey has only short-range sensors but can go twice as fast. Each robot is tested against the best individuals of the previous 5 generations for 40 seconds. The fitness for the predator was inversely proportional to the time needed to catch the prey, the fitness of the prey instead was proportional to the amount of time it managed to escape the predator. In about 25 generations (a few days of co-evolution on the real robots) we observe good chasing and escaping strategies, such as the one shown in this video sequence. If you observe the two robots for further generations, you can notice a variety of different behaviors. For example, in those generations where the prey moves very fast along the walls, the predator does not attempt to follow it but it simply backs up to a wall and there it waits for the prey that travels too fast to avoid it. We can run co-evolutionary experiments thanks to a set of triple rotating contacts. After 20 generations, the predators developed the ability to search for the prey and follow it while the prey escaped moving all around the arena. However, since the prey could go faster than the predator, this strategy did not always pay off for predators. 25 generations later we noticed that predators watched the prey from far and eventually attacked it anticipating its trajectory. As a consequence, the prey began to move so fast along the walls that often predators missed the prey and crashed into the wall. Again, 25 generations later we discovered that predators developed a "spider strategy". Instead of attempting to go after the prey, they quietly moved towards a wall and waited there for the prey which moved so fast that could not detect the predator early enough to avoid it!
Two robots, a predator and a prey, are co-evolved within a square arena. The predator has a short-range sensors (1 cm) and a vision system, whereas the prey has only short-range sensors but can go twice as fast. Each robot is tested against the best individuals of the previous 5 generations for 40 seconds. The fitness for the predator was inversely proportional to the time needed to catch the prey, the fitness of the prey instead was proportional to the amount of time it managed to escape the predator. In about 25 generations (a few days of co-evolution on the real robots) we observe good chasing and escaping strategies, such as the one shown in this video sequence. If you observe the two robots for further generations, you can notice a variety of different behaviors. For example, in those generations where the prey moves very fast along the walls, the predator does not attempt to follow it but it simply backs up to a wall and there it waits for the prey that travels too fast to avoid it. We can run co-evolutionary experiments thanks to a set of triple rotating contacts. After 20 generations, the predators developed the ability to search for the prey and follow it while the prey escaped moving all around the arena. However, since the prey could go faster than the predator, this strategy did not always pay off for predators. 25 generations later we noticed that predators watched the prey from far and eventually attacked it anticipating its trajectory. As a consequence, the prey began to move so fast along the walls that often predators missed the prey and crashed into the wall. Again, 25 generations later we discovered that predators developed a "spider strategy". Instead of attempting to go after the prey, they quietly moved towards a wall and waited there for the prey which moved so fast that could not detect the predator early enough to avoid it!
Two robots, a predator and a prey, are co-evolved within a square arena. The predator has a short-range sensors (1 cm) and a vision system, whereas the prey has only short-range sensors but can go twice as fast. Each robot is tested against the best individuals of the previous 5 generations for 40 seconds. The fitness for the predator was inversely proportional to the time needed to catch the prey, the fitness of the prey instead was proportional to the amount of time it managed to escape the predator. In about 25 generations (a few days of co-evolution on the real robots) we observe good chasing and escaping strategies, such as the one shown in this video sequence. If you observe the two robots for further generations, you can notice a variety of different behaviors. For example, in those generations where the prey moves very fast along the walls, the predator does not attempt to follow it but it simply backs up to a wall and there it waits for the prey that travels too fast to avoid it. We can run co-evolutionary experiments thanks to a set of triple rotating contacts. After 20 generations, the predators developed the ability to search for the prey and follow it while the prey escaped moving all around the arena. However, since the prey could go faster than the predator, this strategy did not always pay off for predators. 25 generations later we noticed that predators watched the prey from far and eventually attacked it anticipating its trajectory. As a consequence, the prey began to move so fast along the walls that often predators missed the prey and crashed into the wall. Again, 25 generations later we discovered that predators developed a "spider strategy". Instead of attempting to go after the prey, they quietly moved towards a wall and waited there for the prey which moved so fast that could not detect the predator early enough to avoid it!
Two robots, a predator and a prey, are co-evolved within a square arena. The predator has a short-range sensors (1 cm) and a vision system, whereas the prey has only short-range sensors but can go twice as fast. Each robot is tested against the best individuals of the previous 5 generations for 40 seconds. The fitness for the predator was inversely proportional to the time needed to catch the prey, the fitness of the prey instead was proportional to the amount of time it managed to escape the predator. In about 25 generations (a few days of co-evolution on the real robots) we observe good chasing and escaping strategies, such as the one shown in this video sequence. If you observe the two robots for further generations, you can notice a variety of different behaviors. For example, in those generations where the prey moves very fast along the walls, the predator does not attempt to follow it but it simply backs up to a wall and there it waits for the prey that travels too fast to avoid it. We can run co-evolutionary experiments thanks to a set of triple rotating contacts. After 20 generations, the predators developed the ability to search for the prey and follow it while the prey escaped moving all around the arena. However, since the prey could go faster than the predator, this strategy did not always pay off for predators. 25 generations later we noticed that predators watched the prey from far and eventually attacked it anticipating its trajectory. As a consequence, the prey began to move so fast along the walls that often predators missed the prey and crashed into the wall. Again, 25 generations later we discovered that predators developed a "spider strategy". Instead of attempting to go after the prey, they quietly moved towards a wall and waited there for the prey which moved so fast that could not detect the predator early enough to avoid it!