The march of the machines
RELATED
Parts of human bodies are becoming machines
and machines are becoming smart, like humans. And we are getting closer
to understanding that ultimate human machine - the brain - by creating a
copy of it on computers. 2013 saw dramatic strides in this
mind-boggling adventure, but we're still not fully there. Here's a quick
look at that and other advancements made this year.
Soft robotics
Forget about those clunky Star Wars-type robots. On the horizon are "soft robots", machines that are made of soft, humanlike materials. And it's not just that they look and behave more like humans. "When our hand wraps around a coffee mug, sensors in the hand inform the brain about the mug's shape, weight, temperature, texture, and the brain informs the muscles what to do about them. That's what we are aiming for," Rolf Pfiefer of the University of Zurich told TOI. His lab has developed the 4-ft tall and cute Roboy, unveiled this March, which has synthetic muscles, tendons and bones. It can walk and even ride a small bicycle. But the best news is this — the Roboy design will be put in the public domain once finalized. Just download and use a 3-D printer to make one for yourself.
Artificial limbs that can feel
In 2012, Zac Vawter walked up 103 floors of Chicago's Willis Tower wearing a mind-controlled bionic leg. But that was only half the battle won because the leg didn't tell the brain what it was feeling. The much vaunted "brain-machine interface" (BMI) was still a one-way street. That changed in 2013.
After Igor Spetic of Ohio lost his hand in an industrial accident, Dustin Tyler of Case Western Reserve University made an artificial hand and connected its wiring to nerve nodes on Spetic's upper arm. Spetic now has a sense of touch through 20 sensors in the prosthesis while his brain is able to control the hand's movements. In an experiment, Spetic was able to pluck the stems of cherries, taking care not to crush the fruit - an ability provided by the sensors.
Others like neuroscientist Miguel Nicolelis of Duke University have implanted tiny electrodes in monkeys' brains so that they can simultaneously move both their artificial arms to carry out given tasks. Nicolelis imaged the brain's cells firing up in tandem with the activity - a glimpse of living brain-machine interaction.
Deep learning
Remember Google's neural network last year that could recognize YouTube cats? Its developer Andrew Ng of Stanford's Artificial Intelligence Lab has created another, more than six times bigger.
Artificial neural networks are virtual models of biological brains. They show signs of deep learning or what Boston University's Stephen Grossberg, a pioneer in the field, calls autonomous intelligence — developing intelligence by interacting with the world.
Deep learning was utilized in several tasks by major players including Microsoft and Facebook, apart from Google. Microsoft used it to translate spoken English into spoken Chinese while Facebook hired top experts in the field to develop more killer apps. "It's both a software and a hardware problem together; the way you scale these networks requires very deep integration of the two," explained Srinivas Narayanan of Facebook to the MIT Technology Review.
Are we on the threshold of a scary sentient machine revolution? Not quite. Remember, the human brain has some 100 trillion connections, machines are struggling at 11 billion.
Decoding the human brain
Humans are making a decisive bid to understand the last frontier — the human brain. Two mega projects were launched this year. The EU-backed Human Brain Project (HBP), aimed at creating a complete computer model of the brain, and the USbacked BRAIN Initiative to image and map all brain cells (neurons ) and their interconnected circuits. Both involve developing new technologies, and both will lead to finally understanding how the 450-gm human brain can do so much. Heidelberg University's physicist Karlheinz Meier, HBP co-coordinator, told TOI that one of the aims is to shift present computing systems, which are very energy and space-intensive, to neuromorphic (" brain-like") machines. "This project is not about developing artificial intelligence. It is about understanding, and then simulating the brain so that we can draw benefits for medicine, computing and of course neuroscience," he said.
Internet of things
The most-sold personal care gizmo by Amazon this holiday season was the Fitbit Wristband, a sleep, exercise and health tracker that straps onto one's wrist. Meanwhile, LG has announced a new line of smart appliances like ovens, vacuum cleaners, and washing machines that will be connected to you through a messaging app. While you work at office, these appliances will be doing their work, updating you, and ready to take instructions from you.
The Internet of Things has come into its own this year. Wearable smart devices, like fitness trackers, watches, glasses, belts etc are most common. Then there are car and driving-related aids - pollution detectors, efficiency analysers and so on. At a bigger level, there are smart traffic lights. And of course, driverless cars made great progress in 2013. The biggest splash was made by Google Chauffeur, a software installed in cars that takes over the car confidently on freeways but is still slightly unsure in residential neighborhoods. University of Parma's BRAiVE moves autonomously on a mixed traffic route open to public traffic.
So what's coming in the future? The key of intelligence will be found in the smart interplay between mind, body, and environment, Christian Freksa, professor of cognitive systems at the University of Bremen in Germany, told TOI. "Future systems will need to have 'social competence', that is, competence for co-operation with other, only partly predictable cognitive agents," he said.
Soft robotics
Forget about those clunky Star Wars-type robots. On the horizon are "soft robots", machines that are made of soft, humanlike materials. And it's not just that they look and behave more like humans. "When our hand wraps around a coffee mug, sensors in the hand inform the brain about the mug's shape, weight, temperature, texture, and the brain informs the muscles what to do about them. That's what we are aiming for," Rolf Pfiefer of the University of Zurich told TOI. His lab has developed the 4-ft tall and cute Roboy, unveiled this March, which has synthetic muscles, tendons and bones. It can walk and even ride a small bicycle. But the best news is this — the Roboy design will be put in the public domain once finalized. Just download and use a 3-D printer to make one for yourself.
Artificial limbs that can feel
In 2012, Zac Vawter walked up 103 floors of Chicago's Willis Tower wearing a mind-controlled bionic leg. But that was only half the battle won because the leg didn't tell the brain what it was feeling. The much vaunted "brain-machine interface" (BMI) was still a one-way street. That changed in 2013.
After Igor Spetic of Ohio lost his hand in an industrial accident, Dustin Tyler of Case Western Reserve University made an artificial hand and connected its wiring to nerve nodes on Spetic's upper arm. Spetic now has a sense of touch through 20 sensors in the prosthesis while his brain is able to control the hand's movements. In an experiment, Spetic was able to pluck the stems of cherries, taking care not to crush the fruit - an ability provided by the sensors.
Others like neuroscientist Miguel Nicolelis of Duke University have implanted tiny electrodes in monkeys' brains so that they can simultaneously move both their artificial arms to carry out given tasks. Nicolelis imaged the brain's cells firing up in tandem with the activity - a glimpse of living brain-machine interaction.
Deep learning
Remember Google's neural network last year that could recognize YouTube cats? Its developer Andrew Ng of Stanford's Artificial Intelligence Lab has created another, more than six times bigger.
Artificial neural networks are virtual models of biological brains. They show signs of deep learning or what Boston University's Stephen Grossberg, a pioneer in the field, calls autonomous intelligence — developing intelligence by interacting with the world.
Deep learning was utilized in several tasks by major players including Microsoft and Facebook, apart from Google. Microsoft used it to translate spoken English into spoken Chinese while Facebook hired top experts in the field to develop more killer apps. "It's both a software and a hardware problem together; the way you scale these networks requires very deep integration of the two," explained Srinivas Narayanan of Facebook to the MIT Technology Review.
Are we on the threshold of a scary sentient machine revolution? Not quite. Remember, the human brain has some 100 trillion connections, machines are struggling at 11 billion.
Decoding the human brain
Humans are making a decisive bid to understand the last frontier — the human brain. Two mega projects were launched this year. The EU-backed Human Brain Project (HBP), aimed at creating a complete computer model of the brain, and the USbacked BRAIN Initiative to image and map all brain cells (neurons ) and their interconnected circuits. Both involve developing new technologies, and both will lead to finally understanding how the 450-gm human brain can do so much. Heidelberg University's physicist Karlheinz Meier, HBP co-coordinator, told TOI that one of the aims is to shift present computing systems, which are very energy and space-intensive, to neuromorphic (" brain-like") machines. "This project is not about developing artificial intelligence. It is about understanding, and then simulating the brain so that we can draw benefits for medicine, computing and of course neuroscience," he said.
Internet of things
The most-sold personal care gizmo by Amazon this holiday season was the Fitbit Wristband, a sleep, exercise and health tracker that straps onto one's wrist. Meanwhile, LG has announced a new line of smart appliances like ovens, vacuum cleaners, and washing machines that will be connected to you through a messaging app. While you work at office, these appliances will be doing their work, updating you, and ready to take instructions from you.
The Internet of Things has come into its own this year. Wearable smart devices, like fitness trackers, watches, glasses, belts etc are most common. Then there are car and driving-related aids - pollution detectors, efficiency analysers and so on. At a bigger level, there are smart traffic lights. And of course, driverless cars made great progress in 2013. The biggest splash was made by Google Chauffeur, a software installed in cars that takes over the car confidently on freeways but is still slightly unsure in residential neighborhoods. University of Parma's BRAiVE moves autonomously on a mixed traffic route open to public traffic.
So what's coming in the future? The key of intelligence will be found in the smart interplay between mind, body, and environment, Christian Freksa, professor of cognitive systems at the University of Bremen in Germany, told TOI. "Future systems will need to have 'social competence', that is, competence for co-operation with other, only partly predictable cognitive agents," he said.
