Asada M. K. F. MacDorman, H. Ishiguro, Y. Kuniyoshi. (2017). Cognitive developmental robotics as a new paradigm for the design of humanoid robots.: Difference between revisions
TUe\20181759 (talk | contribs) No edit summary |
TUe\20181759 (talk | contribs) No edit summary |
||
| (One intermediate revision by the same user not shown) | |||
| Line 3: | Line 3: | ||
As opposed to robots in science fiction, robotics as we know it today has yet to develop the ability to communicate with us and perform a variety of complex tasks in the real world. Thus we advocate the need for cognitive developmental robotics (CDR), which aims to understand the cognitive developmental process that an intelligent robot would require and how to realize them in a physical entity. CDR aims at a constructive approach to realize a mechanism that can adapt to complicated and dynamic changes in the environment based on its capacity for interaction. | As opposed to robots in science fiction, robotics as we know it today has yet to develop the ability to communicate with us and perform a variety of complex tasks in the real world. Thus we advocate the need for cognitive developmental robotics (CDR), which aims to understand the cognitive developmental process that an intelligent robot would require and how to realize them in a physical entity. CDR aims at a constructive approach to realize a mechanism that can adapt to complicated and dynamic changes in the environment based on its capacity for interaction. | ||
We argue that a physical embodiment that allows interaction with the environment is required for an intelligent structure. The significance of embodiment can be concluded from the following: | We argue that a physical embodiment that allows interaction with the environment is required for an intelligent structure. The significance of embodiment can be concluded from the following: | ||
1) Perception and action are not separable but tightly coupled | 1) Perception and action are not separable but tightly coupled | ||
2) Under resource-bounded conditions, an agent is able to learn a sensorimotor mapping from experience (interaction with environment) | 2) Under resource-bounded conditions, an agent is able to learn a sensorimotor mapping from experience (interaction with environment) | ||
3) As the complexity of its task or the environment increases, the agent is able to adapt itself to these changes by learning from the consequences of its actions and adapting this knowledge to new situations. | 3) As the complexity of its task or the environment increases, the agent is able to adapt itself to these changes by learning from the consequences of its actions and adapting this knowledge to new situations. | ||
CDR as a design principle has two sides: The design of a self-developing structure inside the robot’s brain (1) and how to set up the environment so that the robots embedded therein can gradually adapt themselves to more complex tasks in more dynamic situations (2). | CDR as a design principle has two sides: The design of a self-developing structure inside the robot’s brain (1) and how to set up the environment so that the robots embedded therein can gradually adapt themselves to more complex tasks in more dynamic situations (2). | ||
1) Reinforcement learning, which maps from sensory information to actuator outputs. | 1) Reinforcement learning, which maps from sensory information to actuator outputs. | ||
2) Environmental factors include all stimuli from outside the robot. How other active agents respond to the robot is key. If a robot develops expectations concerning how self-induced movements transform sensory projections, passive agents can be detected and modelled from correlations among violated expectations. In this way, the robot develops second-order expectations that may scaffold even more abstract learning in a similar manner. | 2) Environmental factors include all stimuli from outside the robot. How other active agents respond to the robot is key. If a robot develops expectations concerning how self-induced movements transform sensory projections, passive agents can be detected and modelled from correlations among violated expectations. In this way, the robot develops second-order expectations that may scaffold even more abstract learning in a similar manner. | ||
Latest revision as of 13:00, 22 March 2021
Summary
As opposed to robots in science fiction, robotics as we know it today has yet to develop the ability to communicate with us and perform a variety of complex tasks in the real world. Thus we advocate the need for cognitive developmental robotics (CDR), which aims to understand the cognitive developmental process that an intelligent robot would require and how to realize them in a physical entity. CDR aims at a constructive approach to realize a mechanism that can adapt to complicated and dynamic changes in the environment based on its capacity for interaction. We argue that a physical embodiment that allows interaction with the environment is required for an intelligent structure. The significance of embodiment can be concluded from the following:
1) Perception and action are not separable but tightly coupled
2) Under resource-bounded conditions, an agent is able to learn a sensorimotor mapping from experience (interaction with environment)
3) As the complexity of its task or the environment increases, the agent is able to adapt itself to these changes by learning from the consequences of its actions and adapting this knowledge to new situations.
CDR as a design principle has two sides: The design of a self-developing structure inside the robot’s brain (1) and how to set up the environment so that the robots embedded therein can gradually adapt themselves to more complex tasks in more dynamic situations (2).
1) Reinforcement learning, which maps from sensory information to actuator outputs.
2) Environmental factors include all stimuli from outside the robot. How other active agents respond to the robot is key. If a robot develops expectations concerning how self-induced movements transform sensory projections, passive agents can be detected and modelled from correlations among violated expectations. In this way, the robot develops second-order expectations that may scaffold even more abstract learning in a similar manner.