ScienceDaily (Feb. 21, 2005) — Robots that act like rat pups can tell us something about the behavior of both, according to UC Davis researchers.
This robot was designed with the same basic senses and motor skills as a rat pup. (Sanjay Joshi/UC Davis photo).HOW IT WORKS: Most robotic vehicles designed to navigate dangerous terrain rely on artificial vision or sonar systems to find the safest path. But robotic "eyes" don't operate well in low light and sonar can be confused by polished surfaces. The Johns Hopkins University scientists have turned to touch, inspired by how bugs use this sense to navigate dark rooms with varied surfaces. Just like a cockroach's antenna, the artificial version sends signals to the electronic brain of a wheeled robot, enabling the machine to scurry along walls, turn corners, and avoid obstacles in its path.
The antenna is made of cast urethane, a flexible substance that resembles rubber, encased in a clear plastic sheath. It contains six strain gauges, sensors that change resistance as they are bent. The device has been calibrated so that certain electric voltages correspond to certain bending angles as the antenna touches the wall or some other object. This data is fed to the robot's controller, enabling it to sense its position in relation to the way and to maneuver around obstacles. For instance, when the antenna signals that the robot is moving too close to a wall, the controller steers it away.
WHAT IS SWARM INTELLIGENCE: Building "swarms," of robotic insects that work together to adapt to their environment is part of "evolutionary robotics": creating machines that are digitally "bred" to evolve themselves. Swarm intelligence is the notion that complex behavior can arise from large numbers of individual agents each following very simple rules. For example, ants follow the strongest pheronome trail left by other ants to find the most efficient route to a food source, through a process of trial and error. A chunk of the plot in Michael Crichton's novel Prey was inspired in part by an experiment in which a fleet of robotic predators were programmed to seek out "prey" to get their next energy boost. The mechanical "prey," in contrast, were programmed to "graze" on special light sources and to keep alert for potential predators. The respective robots evolved increasingly complex hunting and escape strategies as the swarms of robots accumulated more and more data (in the form of experience) on which to base their decisions.
sciencedaily.com for more info.
Researchers have designed a robot that uses a novel form of electrically activated adhesion to enable it to scale any kind of vertical surface. The robot can even climb surfaces that are dusty or wet, be they concrete, glass, or drywall.
"What's really unique about this is the technology, not the robot," says Harsha Prahlad, senior mechanical engineer at SRI International, a nonprofit research organization based in Menlo Park, CA. There are other robots that can climb walls. But these have usually involved using microscopic fibers designed to mimic the function of the hairlike setae that give geckos their remarkable sticking power, Prahlad says.
In contrast, SRI's robot works by inducing electrostatic charges in the surface of a wall. The advantage here is that the adhesive climbing surfaces of the robot can be turned off, making movement much simpler, says Prahlad. It also makes the robot's adhesive surfaces self-cleaning, he says, thereby avoiding any gradual buildup of dust and dirt that would ultimately reduce the adhesion.
Tests have shown that the robot is capable of generating 1.5 newtons of sticking force per centimeter square of contact with a wall. Presenting his results at this year's International Conference on Robotics and Automation, in Pasadena, CA, Prahlad showed that the robot was able to scale walls while carrying weights of up to 75 pounds.
"It's an interesting and robust approach," says Metin Sitti, a mechanical engineer at Carnegie Mellon University, in Pittsburgh, who has been working on wall-climbing robots for some time. However, he says, the forces generated are just one-tenth as strong as is currently being seen when the gecko-inspired approach is used.
On the plus side, however, the simplicity of Prahlad's approach should make it easier to apply to human wall-climbing applications, says Nicola Pugno, a professor of structural mechanics at Turin Polytechnique, in Italy, who has been working on a sort of Spiderman suit using nanotube-covered adhesive surfaces.
"There is no fundamental reason why you can't scale this up to, say, 200 pounds," says Prahlad. So with a suitable interface, it should be possible to allow a human to use this technology to climb walls, he says. However, such a system would require large pads to increase the surface contact of a person's hands. Otherwise, there would not be enough sticking power to support his or her weight, says Prahlad.
The attractive forces that create the adhesion come from electric fields generated by positive and negative electrodes within the surface pads of the robot, says Prahlad. When a high voltage is applied to these electrodes, positive and negative charges build up, which, in turn, attracts opposite charges from the surface of a wall near the electrodes
IR emitter and IR phototransistor
So, you've fitted some motors to your robot and its happily driving around but it probably keeps colliding with obstacles and getting stuck. You need a way for your robot to detect collisions and move around objects. Enter the humble bump sensor:
A bump sensor is probably one of the easiest ways of letting your robot know it's collided with something. The simplest way to do this is to fix a micro switch to the front of your robot in a way so that when it collides the switch will get pushed in, making an electrical connection. Normally the switch will be held open by an internal spring.
Micro switches are easy to connect to micro controllers because they are either off or on, making them digital. All micro controllers are digital, so this is a match made in heaven. Micro switch 'bump' sensors are easily connected to the Robocore, simply plug them into any free digital socket and away you go.
The following diagram shows a typical circuit for a micro switch bump sensor. The resistor is important because it holds the signal line at ground while the switch is off. Without it the signal line is effectively 'floating' because there is nothing connected to it, and may cause unreliable readings as the processor tries to decide if the line is on or off.
