Mar 31 2010
The Technische Universitaet Muenchen (TUM) researchers have developed an experimental underwater vehicle, known as the ‘Snookie’ underwater robot, that incorporates an artificial sensory organ.
CoTeSys Human Robot Interaction Research
This vehicle was built under the Cognition for Technical Systems (CoTeSys) excellence cluster framework. These researchers had developed the organ based on inspiration from the lateral-line system present in some amphibians and fish. Researches anticipate that such features will empower underwater robots to function autonomously in operations such as the inspection of sewer pipes and deep ocean exploration.
Traditional robots are able to perform their tasks only if there are programmed accurately for taking every step, though they are sturdy and operate in hostile environments, dirt and disease, low light levels, corrosive and toxic gases conditions that are unbearable for humans. On the contrary, autonomous robots will react intelligently with their environment and carry out their functions almost independently in the future. These robots would depend on their sensory perceptions, and would not require rigid programming. This is the only mode wherein such robots will be able to identify their environment and carry out their functions. Yet, senses fail these autonomous robots often due to high temperatures, water, dust, and fumes conditions. This calls for new types of senses that could be sensory type of organs that humans do not have.
The Munich, Germany-based CoTeSys cluster is focusing on developing a technology for mastering such new types of senses. TUM’s biophysicist Prof. Leo van Hemmen expects that the animal domain will offer the methodology to help robots in perceiving their environment. For instance, frogs, scorpions, and fish are able to perceive things undetected by human organs. They have the ability to identify even small vibrations and pressure differences for recognizing threats. Through these senses they are able to exactly visualize their surroundings. This ability helps them to take decisions, for instance, the best way to hide behind defensive obstacles or the best method to catch their prey. The research team is studying the algorithms through which animal brains are able to record their surroundings and are accordingly creating hardware and associated computer programs for helping robots to imitate these animals.
Amphibians and fishes, for instance, incorporate the lateral line that is not possessed by land animals. This sensory organ that extends on both sides of their bodies helps them to perceive even small changes in current and pressure flow. This ability helps them visualize a detailed image of their nearby surroundings at a range equal to their body range, even in murky water conditions. They know where dangers are lurking, location of obstacles, and where they could locate their prey. Lateral lines incorporate many fine sensory hairs positioned in miniature ducts beneath their skin and can even detect minute flow velocity changes. For instance, the African clawed dog, known as Xenopus laevis, has the ability to differentiate between inedible and edible insects based on the detection of water-borne vibrations. In terms of accuracy, such sensors can be compared to the human ear that uses many fine sensory hairs to differentiate between sounds that range from the symphony to the sigh of a wind.
The Munich researchers had analyzed the Astyanax, a blind Mexican cave fish, as a favorite example. The eyes of this fish degenerates as it matures, since a cave inhabitant would not require light during darkness. Yet, this fish has no problem in navigating through its pitch-dark surroundings and flexibly reacts to small variations and adapts in quick time to new surroundings. In fact, the ability of robots to learn to perform similar functions is shown by an underwater robot named ‘Snookie,’ created by an interdisciplinary group of technical specialists and scientists headed by Professor van Hemmen. Snookie was so named after a species having a perch and a distinctive lateral line. This aluminum and Plexiglas based fish has a diameter of 30 cm and length of 80 cm. This fish is filled to its gills with a power supply and an electronic control system. A yellow nose of hemispheric shape that secures the sensors for guiding the underwater vehicle and six propeller gondolas used for driving and positioning the robot, form the key features of the robot..
An underwater vehicle was deliberately selected by the TUM scientists for testing their technology, since these vehicles have to undergo a particular type of challenge that is not encountered by land-based autonomous robots. Underwater visibility is a mere few centimeters. Infrared (IR) detectors usually utilized by land-based robots along with cameras for identifying the surroundings do not function under water. Bad propagation results in limited wireless communication under water. Since energy supplies are based on battery capacity, all underwater systems have to function with utmost accuracy. Underwater robots require maximum reliability, since when something is wrong they will be lost.
Electrical engineer Stefan Sosnowski from the Department of Robotics that is headed by Professor Sandra Hirche is responsible for the underwater craft design. Sosnowski explained that an underwater robot has to function independently like a vehicle on Mars. Biophysicist Dr. Jan-Moritz Franosch, his colleague, along with a student group, has made an artificial lateral line that helps the Snookie robot in detection of movements and obstacles at a distance of the robot’s hand on either side of the robot and in front of the robot. This artificial organ is able to measure flow and pressure variations around the robot using thermistors, instead of traditional, expensive, large, and inaccurate dynamic indicators. A variation in flow velocity results in corresponding variation of heat dispersed in a heated wire, which in turn can be electronically measured by sensor elements quickly in a small space. Using an amount of electrical energy at one tenth second intervals these sensors record fluctuations of pressure. The pressure fluctuation is less than one percent and encompasses an area of only a few square millimeters.
Van Hemmen is convinced that the installation of additional cameras for providing more images would result in very little benefit to robotic intelligence. He feels that it is more critical for robots to perceive different facets of their surroundings using a range of sensors. However, when it relates to the integration of these perceptions, researchers have to focus on intricacies of brain research. One has to analyze the manner in which animals are able to sift through a data mass for filtering out relevant issues and how humans will be able to manage this. He believes that the CoTeSys excellence cluster will offer an opportunity that will, besides answering such queries, will use interdisciplinary collaboration among engineers, information technologists, and physiologists for transferring the newly developed principles to the technological world. He added that alertness implies the reduction of data to only the essentials. He insisted that robots also will have to learn this function, even while faced with a broad range of sensor information.
CoTeSys specializes in this type of interdisciplinary collaboration. This research cluster has combined about 100 scientists who work in very different domain in five research institutions and universities in the Munich region, so that technical systems with improved cognitive capabilities could be developed. They aim to render robots more self-dependent, with ability to analyze and respond flexibly to their environment that involves identifying their environment for independent performance of their allotted tasks. The state and federal governments have earmarked €28 million for funding the joint project that is coordinated by the TUM.