A distributed and morphology-independent strategy for adaptive locomotion in self-reconfigurable modular robots

In this paper, we present a distributed reinforcement learning strategy for morphology-independent life-long gait learning for modular robots. All modules run identical controllers that locally and independently optimize their action selection based on the robot’s velocity as a global, shared reward signal. We evaluate the strategy experimentally mainly on simulated, but also on physical, modular robots. We find that the strategy: (i) for six of seven configurations (3–12 modules) converge in 96% of the trials to the best known action-based gaits within 15 min, on average, (ii) can be transferred to physical robots with a comparable performance, (iii) can be applied to learn simple gait control tables for both M-TRAN and ATRON robots, (iv) enables an 8-module robot to adapt to faults and changes in its morphology, and (v) can learn gaits for up to 60 module robots but a divergence effect becomes substantial from 20–30 modules. These experiments demonstrate the advantages of a distributed learning strategy for modular robots, such as simplicity in implementation, low resource requirements, morphology independence, reconfigurability, and fault tolerance.     

Long-term hearing results following vestibular surgery in Meniere's disease

OBJECTIVES: The purpose of this study is to evaluate the long-term hearing changes following vestibular surgery in patients with Meniere's disease. Study DESIGN: This is a retrospective analysis of patients operated on in a tertiary referral center setting. METHODS: Preoperative, postoperative, and 3- to 9-year postoperative audiograms were analyzed in two patient groups. Twenty-one patients underwent posterior fossa vestibular neurectomy (VN) and five, mastoid endolymphatic sac decompression and shunt (ELS). All frequencies, four-frequency pure-tone averages (PTAs), spondee thresholds, and speech recognition scores were compared for operated ear against nonoperated ear of VN subjects. The results were subjected to a covariance analysis. VN and ELS patients whose hearing deteriorated from "serviceable" (PTA < or =70 dB hearing level) and speech recognition > or =30%) to nonserviceable status were compared using nonparametric statistics. RESULTS: Progressive hearing loss beyond the rate of change of the normal contralateral ear was evident in all patients. Serviceable hearing dropped from 81% to 43% of patients an average of 4 years following VN. CONCLUSIONS: VN patients have significant hearing deterioration over time in the operated ear. This finding suggests that continued postoperative medical management is necessary for patients undergoing VN.     

Tn5 in vitro transposition

This communication reports the development of an efficient in vitro transposition system for Tn5. A key component of this system was the use of hyperactive mutant transposase. The inactivity of wild type transposase is likely to be related to the low frequency of in vivo transposition. The in vitro experiments demonstrate the following: the only required macromolecules for most of the steps in Tn5 transposition are the transposase, the specific 19-bp Tn5 end sequences, and target DNA; transposase may not be able to self-dissociate from product DNAs; Tn5 transposes by a conservative "cut and paste" mechanism; and Tn5 release from the donor backbone involves precise cleavage of both 3' and 5' strands at the ends of the specific end sequences.     

Personality and the inheritance of smoking behavior: a genetic perspective

In contrast to the extensive research effort to understand the genetic contribution to alcoholism risk, there has been little research directed at understanding genetic influences on smoking behavior. Data from large twin studies in Scandinavia and Australia are consistent with a major genetic influence on the probability that an individual will become a smoker ("initiation") and will persist in the smoking habit once smoking has started ("persistence"). We use data from the 1988/1989 follow-up survey of the Australian NH&MRC twin panel to determine to what degree personality measures (Tridimensional Personality Questionnaire, Eysenck Personality Questionnaire--Revised) and attitudinal and sociodemographic variables (social and political conservatism, education, religious involvement) might account for genetic or environmental influences on smoking. While we find significant phenotypic associations between these variables and smoking, these are too modest to account for much of the genetic variance. Possible mechanisms by which this genetic variance may arise are discussed.     

Mechanical responses to two-tone distortion products in the apical and basal turns of the mammalian cochlea

Mechanical responses to one- and two-tone acoustic stimuli were recorded from the cochlear partition in the apical turn of the chinchilla cochlea, the basal turn of the guinea pig cochlea, and the hook region of the guinea pig cochlea. The most sensitive or "best" frequencies (BFs) for the sites studied were approximately 500 Hz, 17 kHz, and 30 kHz, respectively. Responses to the cubic difference tone (CDT), 2F1 - F2 (where F1 and F2 are the frequencies of the primary stimuli), were characterized at each site. Responses to the quadratic difference tone (QDT), F2 - F1, were also characterized in the apical turn preparations (QDT responses were too small to measure in the basal cochlea). The observed responses to BF QDTs and CDTs and to BF CDTs at each site appeared similar in many ways; the relative magnitudes of the responses were highest at low-to-moderate sound pressure levels (SPLs), for example, and the absolute magnitudes grew nonmonotonically with increases in the level of either primary (L1 or L2) alone. The peak effective levels of the CDT and QDT responses were also similar, at around -20 dB re L1 and/or L2. In other respects, however, the responses to CDTs and QDTs and to BF CDTs at each site behaved quite differently. At low-to-moderate SPLs, for example, most CDT phase leads decreased with increases in either L1 or L2, whereas most QDT phase leads increased with increasing L1 and varied little with L2. Most CDT responses also varied monotonically with equal-level primaries (i.e., when L1 = L2), whereas most QDT responses varied nonmonotonically. Different responses also varied in different ways when F1 and F2 were varied. Apical turn QDT responses were observed over a very wide F1/F2 range (F1 = 1-12 kHz), but were usually largest for stimuli <2-4 kHz. Apical turn CDT levels decreased (at rates of approximately 40-80 dB/octave) only when the frequency ratio F2/F1 increased beyond approximately 1.4-1.5. In the basal turn and hook regions, the CDT levels depended nonmonotonically on F2/F1 with the eventual rates of decrease being approximately 200 dB/octave. Optimal frequency ratios for the CDT increased from (F2 < 1.1F1) to (F2 approximately 1.2F1) with increasing SPL in the basal turn, but were stable at around F2/F1 approximately 1.05 in the hook region. CDT phase leads tended to increase with increasing F2/F1 in all three regions of the cochlea, particularly at low-to-moderate SPLs. These findings are discussed in relation to previous studies of cochlear mechanics, physiology, and psychophysics.     

Iliac vessel injury: operative physiology related to outcome

BACKGROUND: Fifty-three patients treated at a level I trauma center with iliac vessel injury were studied to determine if body temperature and acid-base status in the operating room predicts outcome. METHODS: Records were reviewed for demographics, mechanism of injury, body temperature, acid-base status, operative management, and outcome. Statistical methods included Student's t test, odds ratio determination, and chi-square analysis to determine statistical significance. RESULTS: Fifty-three patients (47 male, 6 female) sustained 92 iliac vascular injuries (36 arterial, 56 venous). Mortality was 34%, with 72% of deaths due to shock within 24 hours. Physiologic parameters differed significantly between survivors and nonsurvivors. Odds ratio identified six conditions; the number present predicted outcome. CONCLUSIONS: (1) There are significant differences between initial and final operating room temperature and acid-base status in survivors versus nonsurvivors with iliac vessel injury. Conditions for odds ratio can be calculated and correlated with outcome. (2) A patient with two or more conditions should be considered for an abbreviated laparotomy to allow for reversal of "physiologic failure."     

A second-generation cementless hip prosthesis: improved results over the first-generation prosthesis

Forty-two patients who had a Porous Coated Anatomic (PCA) "E" series, second-generation, cementless hip arthroplasty (Howmedica, Rutherford, NJ) were compared with 42 patients who had a first-generation PCA prosthesis. Patients were directly matched for age, sex, diagnosis, weight, Charnley functional status, and duration of follow-up. All of the operations were done by the same two surgeons, who used the same operative approach and the same postoperative rehabilitation plan. All of the patients were followed up for at least 5 years (range, 60 to 76 months). In the "E" series group, there were 41 of 42 (98%) good and excellent clinical results with a mean Harris hip score of 94 points (range, 46 to 100 points); the first-generation group had 34 of 42 (81%) good and excellent clinical results and a mean Harris score of 81 points (range, 42 to 100 points) (P = 0.001). There was one acetabular component revision in the "E" series group (2%), which can be compared with eight revisions (19%) in the first-generation group (P = 0.012). The incidence of femoral radiolucencies was 19% (eight hips) for the "E" series group compared with 50% (21 hips) in the first-generation group (P = 0.009). The radiolucencies in the "E" series group were small, nonprogressive, and confined typically to zone I. We believe that the improvements in design of the "E" series component may account for these differences.     

OpenLogos machine translation: philosophy,model, resources and customization

This paper reviews the OpenLogos rule-based machine translation system, and describes its model architecture as an incremental pipeline process. The paper also describes OpenLogos resources and their customization to specific application domains. One of the key aspects of rule-based machine translation systems intelligence is the symbology employed by these systems in representing natural language internally. The paper offers details about the OpenLogos semantico-syntactic abstract representation language known as SAL. The paper also shows how OpenLogos has addressed classic problems of rule-based machine translation, such as the cognitive complexity and ambiguity encountered in natural language processing, illustrating how SAL helps overcome them in ways distinct from other existing rule-based machine translation systems. The paper illustrates how the intelligence inherent in SAL contributes to translation quality, presenting examples of OpenLogos output of a kind that non-linguistic systems would likely have difficulty emulating. The paper shows the unique manner in which OpenLogos applies the rulebase to the input stream and the kind of results produced that are characteristic of the OpenLogos output. Finally, the paper deals with an important advantage of rule-based machine translation systems, namely, the customization and adaption to application-specific needs with respect to their special terminology and transfer requirements. OpenLogos offers users a set of comfortable customization tools that do not require special knowledge of the system internals. An overview of the possibilities that these tools provide will be presented.     

Estimation of Volumetric Oxygen Concentration in a Marine Environment with an Autonomous Underwater Vehicle

Dissolved oxygen (DO) concentration is a key indicator of the health and productivity of an aquatic ecosystem. This paper presents a new method for high‐resolution characterization of DO as a function of both space and time. The implementation of a new oxygen optode in an Iver2 autonomous underwater vehicle (AUV) is described, which enables the system to measure both absolute oxygen concentration and percentage saturation. Also described are details of AUV missions in Hopavågen Bay, Norway, which consisted of a series of repeated undulating lawnmower patterns that covered the bay. Through offline postprocessing of data, sensor characteristic models were developed, as well as a 3D lattice time series model. The model was constructed by estimating DO at each 3D lattice node location using a 1D Kalman filter that fused local measurements obtained with the AUV. By repeating model construction for several missions that spanned 24 h, estimates of DO as a function of space and time were calculated. Results demonstrated (1) the AUVs ability to repeatedly gather high‐spatial‐resolution data (2) significant spatial and temporal variation in DO in the water body investigated, and (3) that a 3D model of DO provides better estimates of total DO in a volume than extrapolating from only a single 2D plane. Given the importance of oxygen within an ecosystem, this new method of estimating the quantity of DO per volume has the potential to become a reliable test for the health of an underwater ecosystem. Also, it can be refined for detecting and monitoring a range of soluble gases and dispersed particles in aquatic environments, such as dissolved O₂ and CO₂ around production facilities such as fish farms, or dispersed hydrocarbons and other pollutants in fragile ecosystems. © 2012 Wiley Periodicals, Inc.