Robot Builders

The Robotics Institute released a report recently titled Parrots: A Range Measuring Sensor Network (PDF format) that describes a wireless sensor network that can be used for localization by indoor mobile robots. The network consists of nodes, called Parrots, that measure the range to other nodes using ultrasonic sensors. Each node includes a 16bit microcontroller, a wireless link, and a sensor board with 4 ultrasonic sensor arrays. When lots of Parrots are placed in an area, they automatically form a multi-hop ad-hoc wireless network. The robot carries a Parrot node as well, allowing the network to accurately map the location of the robot as it moves through the area. This research was done by the CMU Field Robotics Center. -more-

Several companies have started to sell power ‘pads’ that can charge your cellphone when you put it on the pad’s surface. But these silicon-based pads are expensive — and relatively ’specialized.’ Now, Japanese researchers have built a plastic sheet which could power all the devices placed close to it. So far, this 4-layer sheet, which uses printed organic transistors and plastic MEMS switches, can deliver up to 40 watts of power — enough for some laptops. The technology is apparently efficient and inexpensive to produce. But as the devices to recharge will need to incorporate a special receiver, don’t expect to see these plastic power sheets on sale before several years. But read more… read story

Elevation maps are a popular data structure for representing the environment of a mobile robot operating outdoors or on not-flat surfaces. Elevation maps store in each cell of a discrete grid the height of the surface at the corresponding place in the environment. However, the use of this 21/2-dimensional representation, is disadvantageous when utilized for mapping with mobile robots operating on the ground, since vertical or overhanging objects cannot be represented appropriately. Furthermore, such objects can lead to registration errors when two elevation maps have to be matched. In this paper, an approach is proposed that allows a mobile robot to deal with vertical and overhanging objects in elevation maps. The approach classifies the points in the environment according to whether they correspond to such objects or not. Also presented is a variant of the ICP algorithm that utilizes the classification of cells during the data association. Additionally, it is shown how the constraints computed by the ICP algorithm can be applied to determine globally consistent alignments. Experiments carried out with a real robot in an outdoor environment demonstrate that the proposed approach yields highly accurate elevation maps even in the case of loops. Experimental results are presented demonstrating that that the proposed classification increases the robustness of the scan matching process.

According to researchers at the University of California at San Diego (UCSD), semiconducting nanowires can be used to build perfect light detectors with single-photon sensitivity. The zinc oxide (ZnO) nanowires they’ve used are ideally suited to develop “new photodetector architectures for sensing, imaging, memory storage, intrachip optical communications and other nanoscale applications.” So far, the engineers have demonstrated in their labs that nanowires are extremely sensitive photodetectors because of their specific geometry which combines large surface areas and small volumes. Now we have to wait for real applications.

The Institute for Ethics and Emerging Technologies has published an essay by Michael Anissimov detailing his reasons for believing that artificial intelligence is dangerous. In short he believes that an intelligent agent with different goals than humans, or which has the capacity to have different goals, is inherently a threat to the survival of the human race. His conclusion appears based on the assumption that only humans can be equipped with “freedom, empathy, self-determination, consensus-building, conflict resolution, aesthetics, camaraderie, and rapport.” Why won’t machines have these qualities? “What programmer has the time to do all that work when general intelligence without the human-like morality will be significantly easier to achieve?”

This paper introduces a systematic constraint-based approach to specify complex tasks of general sensor-based robot systems consisting of rigid links and joints. The approach integrates both instantaneous task specification and estimation of geometric uncertainty in a unified framework. Major components are the use of feature coordinates, defined with respect to object and feature frames, which facilitate the task specification, and the introduction of uncertainty coordinates to model geometric uncertainty. While the focus of the paper is on task specification, an existing velocity- based control scheme is reformulated in terms of these feature and uncertainty coordinates. This control scheme compensates for the effect of time varying uncertainty coordinates. Constraint weighting results in an invariant robot behavior in case of conflicting constraints with heterogeneous units. The approach applies to a large variety of robot systems (mobile robots, multiple robot systems, dynamic human-robot interaction, etc.), various sensor systems, and different robot tasks. Ample simulation and experimental results are presented.

read story

The following motion coordination problem is studied: given n mobile vehicles and n source-destination pairs in the plane, what is the minimum time needed to transfer each vehicle from its source to its destination, avoiding conflicts with other vehicles? In the proposed model, vehicles do not explicitly communicate their intentions, and only have sensory information about the current position and velocity of their neighbors to ensure no conflicts. The environment is free of obstacles and a conflict occurs when the distance between any two vehicles is smaller than a velocity-dependent safety distance. The situation analyzed in which the vehicle size is such that at least a constant fraction of the n vehicles can be fitted inside the environment simultaneously. In the “best” case in which the source and destination points can be chosen ideally to maximize the transfer efficiency, it is shown that the transfer takes ([UNKNOWN][UNKNOWN]n) time to complete, where [UNKNOWN] is the average distance between the source and destination points. It is shown that there exist a “worst” case distribution of the source and destination points, for which the transfer of vehicles takes at least (n) time. The case is also analyzed in which source and destination points are generated randomly according to a uniform distribution, and an algorithm is presented providing a constructive upper bound on the time needed to transfer vehicles from sources to their corresponding destinations, proving that the transfer takes ([UNKNOWN]n) time, with high probability, thus recovering the best case performance.

read story

If researchers from Duke University can successfully install an updated version of a new image-viewing software found on clinical ultrasound scanners, it soon could be possible to see developing fetuses in the womb in living 3-D movies. These 3-D ultrasound imaging systems don’t need any additional hardware, except special goggles. But as new monitors able to directly watch stereo 3-D images are now in development, this need for glasses should also been eliminated.

The BBC reports on a recent cortical simulation experiment run on one of the older IBM BlueGene/L supercomputer prototypes. The goal of the researchers is to create a complete, real-time simulation of a mouse brain. The current run simulated only half of a scaled down mouse brain consisting of 8000 neurons, each with 6300 synapses. Half of a real mouse brain has about eight million neurons, each with 8000 synapses. The IBM BlueGene/L prototype used, probably the DD2, has a relatively modest 4096 processers with 256MB of memory each. The BlueGene/L computers run SUSE SLES 9 Linux and are among the fastest computers on Earth. While the DD2 comes in at only 11 TeraFLOPS, the production model BlueGene/L tops the list at 360 TeraFLOPS. The computer could run a simulation of this complexity with a resolution of 1ms and a neural firing rate of 1Hz, or about 1/10 of realtime. So one second of simulated mouse thoughts took 10 seconds to calculate. The researchers claim to have seen “biologically consistent dynamical properties”, including the spontaneous formation of neuron groups and the staggered, co-ordinated firing of synapses. A few more details can be found in the short project description document, Towards Real-Time, Mouse-Scale Cortical Simulations (PDF format).

Elevation maps are a popular data structure for representing the environment of a mobile robot operating outdoors or on not-flat surfaces. Elevation maps store in each cell of a discrete grid the height of the surface at the corresponding place in the environment. However, the use of this 21/2-dimensional representation, is disadvantageous when utilized for mapping with mobile robots operating on the ground, since vertical or overhanging objects cannot be represented appropriately. Furthermore, such objects can lead to registration errors when two elevation maps have to be matched. In this paper, an approach is proposed that allows a mobile robot to deal with vertical and overhanging objects in elevation maps. The approach classifies the points in the environment according to whether they correspond to such objects or not. Also presented is a variant of the ICP algorithm that utilizes the classification of cells during the data association. Additionally, it is shown how the constraints computed by the ICP algorithm can be applied to determine globally consistent alignments. Experiments carried out with a real robot in an outdoor environment demonstrate that the proposed approach yields highly accurate elevation maps even in the case of loops. Experimental results are presented demonstrating that that the proposed classification increases the robustness of the scan matching process.

read story