MIT Trains Robots To Jump
Nerval's Lobster writes: MIT just announced that its researchers have programmed a robotic cheetah that can leap over obstacles without a prompt from a human controller. The machine's onboard sensors rely on reflected laser-light to judge obstacles' distance and height, and use that data to fuel the algorithm for a safe jump. The robot's controlling algorithm takes into account such factors as the speed needed to launch its mass over the obstacle, the best position for a jump, and the amount of energy required from the onboard electric motor. As of this writing, the robot can clear 90 percent of obstacles on an open track. "A running jump is a truly dynamic behavior," Sangbae Kim, an assistant professor of mechanical engineering at MIT, is quoted as saying in a university press release. "You have to manage balance and energy, and be able to handle impact after landing. Our robot is specifically designed for those highly dynamic behaviors." For years, some tech pundits have worried that robots and software will gradually replace human workers in key industries such as manufacturing and IT administration. Now they have something else to fret over: Robots replacing the world's hurdlers.
You know, somehow I'm betting a significant chunk of Slashdotters can't do a running jump without falling on their ass. ;-)
Lost at C:>. Found at C.
Is there a reason why all the obstacles are flat, low and pink?
Most likely because they already had some pink Styrofoam lying around.
Can the lasers only see pink objects?
Who cares? Obstacle detection with lasers is already a solved problem, and that was NOT the point of this research. The point is the algorithms and mechanism of jumping.
That's because the obstacles are meant to represent people gunned down by the robotic cheetah. Robot Cheetah will need to leap over them to gun down more people. Otherwise, we could just send wave after wave of men at the robot cheetahs until they are blocked in by corpses.
My sci-fi novel, Ghost Thief, is now available from Amazon.com.
If all the obstacles have to be pink to be detected that might be an issue.
Thousands of cars already have laser based obstacle detection that works in real time at highway speed. They work just fine with non-pink objects. They may have used pink in this instance just to make their image analysis easier. The color of the obstacles has no significance whatsoever to the basics of jumping and landing, and it is silly to fixate on that.