Robotics

Classical AI (GOFAI)

Shakey was the first robot that used a layered architecture: complex behaviors are hierarchically organized

limitations: not embedded in real-life environment and can never really come to terms with real-life problems and challenges

can only operate in constrained environment

cannot learn to solve problems - all solutions to problems are built in

dynamical systems

Are used to understand how agents are embedded in their environments

Use calculus-based methods to track the evolving relationship between a small number of variables over time

situated cognition

dynamical systems like approach to robotoics

start small and focus on ecologically valid problems

ex: crickets move their ears in direction of ear which vibrated the most to cricket like sounds

biorobotics

using knowledge of living insects, as well as AI, to create agents capable of moving about and solving problems in their environment

morphological computation

exploiting features of body shape to simplify what might otherwise be highly complex information-processing tasks

subsumption architectures

do not operate by executing algorithims to map their surrondings

they use simple stimulus-response mechanisms

intelligence aggregates from bottom-up

based on idea that intelligence does not require formal symbolic representations

robots constructed with this architecture make reflexive response to environmental stimuli

made up of layers and built up from behaviors

obstacle avoidance layer - subsumption architectures

Directly connects perception (sensing an obstacle) to action (either swerving to avoid the obstacle, or halting when the obstacle is too big to go around)

rodney brook’s robot allen

basic layer is obstacle - avoidance layer

over time more layers are added, mimicing evolution

no central controller like PLANEX in SHAKEY

direct perception-action links allows robot to deliver immediate motor response to sensory input

hybrid architectures have

Subsumption architecture for low-level reactive control (“scaled-up insects”) WITH

Traditional central planner for high-level decision-making (“scaled- down supercomputers”) grafted onto them

situated vs. embodied cognition

situated: “embedded in the world, and which does not deal with abstract descriptions, but through its sensors with the here and now”

embodied: “one that has a physical body and experiences the world directly through the influence of the world on that body”

xenobot

built from ground up using biological cells

made of skin cells and heart cells in the form of stem cells harvested from frog embryo

Designed and programmed by a supercomputer using an evolutionary algorithm

biggest challenges in fine motor movements of humanoid robots

Recent breakthroughs in dexterity have been achieved through imitation learning

A roboticist puppeteers pairs of metallic robotic arms – parallel jaw grippers

El Alcance

undergraduate students at UC Berkeley recently built and

deployed a robotic arm to feed paralyzed patients at non-profit in Chile

reinforcement learning

subdivides each demonstration into a series of sub-tasks and mines successful simulations

The robot then practices the sub-tasks

It is thus able to learn more from less data

affective computing

Computing that relates to, arises from, or deliberately influences emotions

computer could tell when you are tired or frustrated based on your facial expression and recommend a rest break