Techorama 2017

(Ro)Bots are supposed to take over a lot of jobs in the future. In order to get to that point, they need to be able to operate without human assistance. One of the major concerns that pops up in completely autonomous systems is how to deal with failure. The most common response is to assist the software with human intervention. In practice however this does not work, humans are just too easily distracted to count on them in an emergency situation. In this session we'll look at design and implementation patterns that we can leverage in order to let a system heal itself, or at least prevent it from doing any damage if operating incorrectly.