BGU to develop software for robotics challenge

The US government’s Defense Advanced Research Projects Agency (DARPA) is responsible for the development of new technologies for use by the military. The agency at the Department of Defense, which has been responsible for funding the development of many technologies that have had a major effect on the world, also finances a “Robotics Challenge.” Now Ben-Gurion University of the Negev in Beersheba has become the only foreign university chosen to lead a team on a track to develop control software for the DARPA robot.

BGU has been given a $375,000, nine-month grant to develop control software for a disaster-response operations robot in the Israeli part of the project, called Robil.

This technology is expected to improve the performance of robots that operate in the rough terrain and austere conditions characteristic of disasters and use vehicles and tools commonly available in populated areas. The technology will also work in ways easily understood by subject matter experts untrained in the operation of robots, and will be governed by intuitive controls that require little training.

“Natural and man-made disasters have caused suffering for people around the world, in past ages, today, and surely tomorrow. The devastation of disasters such as Fukushima, the Deepwater Horizon oil spill and the Chilean Copiapó mine collapse all serve to highlight our fragility in the presence of unforeseen events,” according to the challenge’s website.

“Often, subject matter experts are available with the knowledge to prevent further damage, yet are unable to get close enough to complete their mission – be it from nuclear contamination, intense pressure, structural instability, or many other threats to human safety. Our best robotic tools are helping, but they are not yet robust enough to function in all environments and perform the basic tasks needed to mitigate a crisis situation. Even in degraded post-disaster situations, the environment is scaled to the human world, requiring navigation of human obstacles such as doors and stairs, manipulation of human objects such as vehicles and power tools, and recognition of common human objects such as levers and valves.”

The software will control the robot developed by Boston Dynamics, Inc., based on its Atlas humanoid robot platform and modified to meet the needs of the DARPA Robotics Challenge. “Robil’s team is an ad-hoc consortium led by BGU comprised of the leaders of the Israeli robotics industry and academia. It includes 20 key personnel and over 40 graduate students and engineers,” says Robil team leader Prof. Hugo Guterman of BGU’s electrical and computer engineering department.

WHY HASTE MAKES WASTE

It was the great 18th century American inventor and statesman Benjamin Franklin who said, “Take time for all things: Great haste makes waste.” But why do our brains make more mistakes when we act quickly? A new study, just published in the prestigious journal Neuron and conducted by Prof. Richard Heitz and Prof. Jeffrey Schall at Vanderbilt University in Tennessee, has found that the brain actually switches into a special mode when pushed to make rapid decisions.

“This is a question that is very basic to our experience as human beings, and something that we encounter on a daily basis,” Heitz said. “If we can understand how our brain changes when we are pushed to respond faster, we have gone a long way toward understanding the decision- making process in general.”

The tradeoff between speed and accuracy is a universal aspect of decision-making that has been studied extensively in humans. However, none of the previous experiments were capable of studying the decision-making process at the level of individual neurons in the brain. Though it’s easy to set up tests that prompt human subjects to switch between speedy and accurate decision-making, the methods for measuring human brain activity do not have the required speed or resolution. On the other hand, it is possible to measure the activity of individual neurons in monkeys – but no one knew how to train them to vary the speed of their decision-making.

The scientists developed a method for teaching monkeys to switch back and forth between fast and accurate decisionmaking in a task that involved picking out a target from an array of objects presented on a computer screen. In one experimental condition, monkeys learned that only accurate responses would be rewarded.

In another condition, they learned that making some mistakes was okay, as long as the decisions were fast. Meanwhile, the researchers monitored signals from single neurons in their prefrontal cortex – the area in the brain dedicated to higher cognition.

“They found that identical information presented to the brain is analyzed differently under speed stress than under accuracy stress,” Schall said. These unexpected results are controversial and important because they are at odds with currently accepted mathematical models of decision- making, which are being used to understand psychiatric and neurological disorders. For example, people with certain types of brain damage seem to get stuck in a hasty, impulsive mode of deciding, and the models provide some indication for how this might happen mechanically in the brain. Heitz and Schall have shown how these mathematical models can be modified to make them consistent with the new results.

So, does haste inevitably make waste? According to Schall, it all depends. “Haste makes waste when a mistake entails dire consequences. But there are many situations in life when the cost of not acting is higher than making an error in judgment.

For example, if the decision is whether or not to shut down a nuclear reactor in the presence of a potential meltdown, I’d prefer haste.”