A group led by Robert Shepherd, assistant professor of mechanical and aerospace engineering and principal investigator of Organic Robotics Lab at Cornell University, have published a paper which reveals their use of stretchable optical waveguides – utilizing absorbency and reflection intensity of light waves to scan and interpret the environment – for use as strain sensors in soft robotics. Robots able to effectively detect strain (such as pressure) will be better able to use tactile sensation to interact more humanly with the world. To show off their advances the group used their photonic strain sensors in a soft robotic hand to conduct experiments with the aim of mimicking a real hand’s capability. In one of the experiments the hand could choose the ripest, hence softest, tomato out of three.
A prototype of a pond skater like robot has been built by a collaboration of university researchers from Bristol, UK. The robot sustainably powers itself by eating, digesting and defecating as it moves over a water surface. The robot ingests food through a polymer mouth. The food then goes through a series of microbial fuel cells, where microbes break it down and convert the energy into electricity. The waste material is ejected out of the rear end as the mouth sucks in another mouthful of water. Currently the energetically autonomous robot is just proof of concept; however, in the future it could have applications in cleaning polluted water or carrying out tasks in other harsh environments. Problems the team face are increasing the amount of energy harvested, and developing solutions for environments with intermittent, sparse food sources.
Scientists at the University of California have built a prototype robot which can jump from a standing start to a height of 1.46 metres. The scientists looked at a Senegal bushbaby, a small primate, which has the highest vertical jump agility (a measure of the ability of a robot or animal to repeatedly and rapidly jump vertical distances from a state of nonaction) of any animal. The paper reports that the scientists managed to create a robot with a vertical jump agility 23% higher than previous attempts by jumping robots. They did this by designing and using a specialised leg mechanism based on, and translated from, the physiology of the bushbaby’s – using the concept of power modulation in which stored energy is transferred into elastic structures (tendons) and released in a burst of power not achievable by muscles alone.
At NASA’s Jet Propulsion Laboratory, Pasadena, USA, a collaboration led by Douglas Hofmann has been investigating bulk metallic glass (BMG) for manufacture of robot gears. BMG is created by changing the atomic structure of a metal through a process of heating and rapid cooling, to trap it in a liquid atomic arrangement, meaning that it has many properties of glass including flowability, and can be injection moulded. A property that makes BMG good for space exploration is that it doesn’t become brittle in extreme cold – testing has demonstrated the gears can work smoothly at temperatures of -200°C.
Hofmann also believes that this material would be beneficial for use in commercial robotics. Hofmann opines that strain wave gears (metal rings that flex as the gears turn) with their wide use in humanoid robots’ joints to prevent arm shake, are expensive to produce, but show no great benefit compared with BMG manufactured strain wave gears.
At NASA’s Jet Propulsion Laboratory, Pasadena, USA, scientists have been demonstrating their prototype POINTER (Precision Outdoor and Indoor Navigation and Tracking for Emergency Responders) system. It uses quasistatic fields (a type of electromagnetic field which moves so slowly that it appears static). Although they have shorter ranges (just a couple of hundred metres), these fields have a benefit in that they can travel through obstacles which would render radio waves useless.
Development continues with the aim of miniaturization – the team have shown a device which weighs 11.7 grams, though they hope to make it small enough to be put in a pocket or clipped to a belt buckle – and commercialization.
The system could assist search and rescue teams, or in military and industrial contexts for tracking robots that are operating underground or in other locations where radio wave location would be impractical.
Researchers at Duke University have discovered a way to speed up a robot’s motion planning process threefold and using one-twentieth of the power of conventional processes.
Motion planning requires the construction of a probabilistic road map (PRM). This consists of points in space that are free of obstacles, with edges (lines) connecting them. The robot can then use the edges to calculate how to move.
The researchers use a specialised processor to precompute (during setup the robot is programmed with its environment’s PRM) and use parallelism (using multiple processors simultaneously) to generate a motion plan. They then looked to see which edges were the most commonly used and pruned those that weren’t used. This is important because reducing the number of edges enables the field-programmable gate array (FPGAl a processor) to process information quicker than previously. Each circuit in the FPGA receives input of the 3D location of a single pixel in a depth image. It then outputs a single bit that shows whether the edge collides with the pixel. If it collides then that edge is removed from the PRM. Hence through a process of elimination the PRM is left with collision free paths.
University of Tokyo’s JSK lab researchers have given their humanoid robot Kengoro the ability to sweat. They manufactured water permeable (spongy) aluminium which they then attached to act as a skeleton to Kengoro. Due to the enhanced cooling effect of the sweat, one cup of water enables Kengoro to do push-ups continuously for up to eleven minutes. Although this design is not as efficient as conventional radiator-based cooling systems, the researchers claim it is better than aircooling or water-filled piping systems.
Israeli robotics company, FFRobotics, has created a robot which can pick fruit. The robot possesses image processing technology and uses algorithms that enable it to locate and differentiate between ripe or unripe, and useable or damaged fruit. In a short video on the company’s website the prototype robot is seen working: its twelve three-fingered arms pick apples – reputedly having a capacity up 10,000/hour – from a tree and drop them into collection baskets. After a row of trees has been picked the robot needs to be driven by a human to another picking location.