Wednesday, 5 April 2017

Of Embodiment, Robots and Nightmares…

Robots are here to stay, whether we like it or not, they have been slowly crawling into the militia, factories, schools and even our homes during the last decade. And it is fascinating how the insights of new approaches to cognition have greatly influenced these intriguing mechanical creatures.

As described by Louise Barret (2016) the famous 4E cognition approach —cognition as embodied, embedded, extended and enacted— has radically changed the job description of the brain. Brains are no longer “representational”, now they can be viewed as “performative”. Meaning that it is not longer their job to model the world, but to guide and control action in a dynamic and unpredictable world. This greatly lightens the load of brains, even if they are made of silicon and some metallic alloy.

In a similar fashion, Rolf Pfeifer (2007) states that if it is properly applied, embodiment can lead to surprising insights. For instance, instead of making the “brain” of the robot —or microprocessor— responsible for controlling the movement of all the limbs and joints, an embodied perspective distributes control and processing to all aspects of the agents (its central nervous system, the material properties of its musculoskeletal system, the sensor morphology and the interaction with the environment).

The old approach usually  yields robots that are clumsy and energy inefficient. The new paradigm, however, is allowing robots to better cope with the changeable environments they encounter. As cited in Barret (2016), Rodney Brooks demonstrated —inspired by the enaction approach to cognition— that flexible, intelligent behaviour does not require a representation-heavy computational model to achieve. In fact, Brooks invented Roomba, a robot that shows that it is not necessary to go to the expense of representing the world when the world itself already contains all the information needed.

Robots are no longer limited to a static environment in a mundane factory, now they are out in the wild. The following videos pretend to show some examples of robots that attempt to adapt to their environment.

Roomba, the hoovering robot that relinquished having a full representation of the world.


Cassie —the human-ostrich bipedal robot— and SpotMini —the dog-like robot— show that the morphology and material properties can truly lighten the load of the “brain” and make the movement feel more natural.



And last but not least, Handle a robot that if fails to show that it can adapt to the environment by alternating between wheels and legs, at least it will succeed in inducing you some nightmares.


References:

Barret, L. (2016). Why Brains Are Not Computers, Why Behaviorism Is Not Satanism, and Why Dolphins Are Not Aquatic Apes. BEHAV ANALYST 39: 9-23.

Pfeifer, R. (2007). Self-Organization, Embodiment, and Biologically Inspired Robotics. Science 318, 1088. 

No comments:

Post a Comment