Perception and learning are the key prerequisites for making robots or embedded
system truly autonomous. These intelligent robots/machines must deal with the
enormous uncertainty that exists in the physical world.

Uncertainty arises from many sources. Robot environments are inherently
unpredictable. The uncertainty is particularly high for robots operating in the
proximity of people. Sensors are limited in what they can perceive. Limitations
arise from the range and resolution of a sensor, noise and sensor failure. Robot
actuation involves motors in which uncertainty arises from effects like control
noise, wear-and-tear and mechanical failure. In addition, the approximate nature
of algorithms causes uncertainty. As robotics is now moving into the open world,
managing uncertainty of perception and learning has become the most important
step towards robust real-world robot systems.

This course will cover modern probabilistic and statistical techniques, relative
new approaches to robotics that pay tribute to the uncertainty in perception and
learning.