Archive | Situational awareness & safety critical RSS feed for this section

What my Eye Tells my Brain about the Frog

18 Mar

frog33

Everything we know about the world comes from our senses. This is true of all our machines, too. Our brains process all of this snowstorm of sensation and crystallize from it a crisp, concise, relevant, and actionable model of the world around us. Our machines are not yet so adept.

Since the publication of Judea Pearl’s Probabilistic Reasoning in Intelligent Systems in 1988, whole academic fields have been reborn and revitalized. Of interest here are the fields of artificial intelligence (AI) and robotics. The value represented by the application of Bayes Networks and causality to the processing of sensor signals cannot be understated. The Google Self Driving Car, the Nest thermostat, and the sensor payloads of the US Government’s Predator and Reaper drones all owe a great debt to Pearl’s seminal work.

Deep networks are also a powerful tool for recognizing patterns, detecting anomalies, and making predictions based upon signals from sensors and actuators embedded in the real world. We will in particular be discussing practical applications of deep networks, and in general the works of giants like Geoffrey Hinton.

The focus will nearly always be on applications. For example we might ask (and work toward solving) an important, nerdy, dirty, “Bayes Networkian” questions like this: when my house’s arc fault interrupter senses that my toaster has shorted out, why on earth isn’t this information shared with my smoke detector and carbon monoxide detector? And my heat-sensor-equipped Roomba, which drives a bit closer to investigate?

We will also cover topics such as multisensor fusion, hyperspectral image sensors, supervised and unsupervised machine learning, hidden Markov models, prognostics, computer vision, robotics and biomimetics. And importantly, where sensing and cognition directly affects people: cognitive science, situational awareness. And sensor augmentation: the enhancement of human perception through the extension of human sensation.

And lastly, I am passionate about the impact of automation on technological and structural unemployment. I tend to be more in the Rodney Brooks (optimist) camp than the Race Against the Machines camp on this key issue, and we will explore ways to make smart automation work for real people.

Advertisements