Reinforcement learning of iterative behaviour with multiple sensors

Reinforcement learning allows an agent to be both reactive and adaptive, but it requires a simple yet consistent representation of the task environment. In robotics this representation is the product of perception. Perception is a powerful simplifying mechanism because it ignores much of the complexity of the world by mapping multiple world states to each of a few representational states. The constraint of consistency conflicts with simplicity, however. A consistent representation distinguishes world states that have distinct utilities, but perception systems with sufficient acuity to do this tend to also make many unnecessary distinctions.In this paper we discuss reinforcement learning and the problem of appropriate perception. We then investigate a method for dealing with the problem, called theLion algorithm [1], and show that it can be used to reduce complexity by decomposing perception. The Lion algorithm does not allow iterative rules to be learned, and we describe modifications that overcome this limitation. We present experimental results that demonstrate their effectiveness in further reducing complexity. Finally, we mention some related research, and conclude with suggestions for further work.     

Theoretical analysis of oxidative coupling of methane and Fischer Tropsch synthesis in two consecutive reactors: Comparison of fixed bed and membrane reactor

In this paper, theoretical performance of Fischer Tropsch (FT) synthesis is analyzed where its feed comes from an oxidative coupling of methane (OCM) reactor. In this model based analysis, two consecutive reactors are intended that first reactor is OCM and second reactor is FT and FT reactor performance is compared in two conditions of fixed bed and membrane reactor (MR). The parameters concerned, were CH₄/O₂ ratio, contact time, temperature, and amount of N₂ in OCM feed. High CH₄/O₂ ratio gave low yield of C₂₊ in OCM due to insufficient oxygen, but favored FT reaction due to more yield of C₅₊ and other products. Therefore, it was concluded that production and yield of C₅₊ could be more by use of these configurations.     

A NEW CLASS OF SIMILARITY SOLUTIONS OF AN UNSTEADY ELECTRICALLY CONDUCTING FREE-FORCED CONVECTIVE FLOW IN A VERTICAL POROUS SURFACE WITH DUFOUR AND SORET EFFECTS

The 2-D unsteady magnetohydrodynamic free-forced convective boundary layer flow of a viscous incompressible fluid is studied numerically taking into account heat and mass transfer. The fluid is subjected to uniform heat and mass fluxes embedded in a porous medium by the presence of coupled Dufour and Soret effects. A new class of similarity equations has been obtained by introducing a time-dependent length scale and a corresponding similarity variable. The resulting equations are then integrated numerically using the Nachtsheim-Swigert shooting iteration technique along with the sixth-order Runge-Kutta integration scheme. By developing locally similar solutions of the fluid flow, the behavior of the velocity, temperature, and concentration fields as well as the rate of heat transfer, wall temperature gradient, rate of mass transfer, and skin friction coefficient have been investigated. The effects of Grashof number (Gr), modified Grashof number (Gm), combined effects of the porous and magnetic parameter (S), suction/injection parameter Fw, Brinkman number (Br), Soret number (Sr), and Dufour number (D_f) have been observed on the flow field and discussed.     

Numerical fitting‐based likelihood calculation to speed up the particle filter

The likelihood calculation of a vast number of particles forms the computational bottleneck for the particle filter in applications where the observation model is complicated, especially when map or image processing is involved. In this paper, a numerical fitting approach is proposed to speed up the particle filter in which the likelihood of particles is analytically inferred/fitted, explicitly or implicitly, based on that of a small number of so‐called fulcrums. It is demonstrated to be of fairly good estimation accuracy when an appropriate fitting function and properly distributed fulcrums are used. The construction of the fitting function and fulcrums are addressed respectively in detail. To avoid intractable multivariate fitting in multi‐dimensional models, a nonparametric kernel density estimator such as the nearest neighbor smoother or the uniform kernel average smoother can be employed for implicit likelihood fitting. Simulations based on a benchmark one‐dimensional model and multi‐dimensional mobile robot localization are provided. Copyright © 2015 John Wiley & Sons, Ltd.     

Occurrence of Campylobacter spp. in raw and ready-to-eat foods and in a Canadian food service operation

The occurrence of Campylobacter spp. in a variety of foods from Ottawa, Ontario, Canada, and raw milk samples from across Canada was determined over a 2-year period. The samples consisted of 55 raw foods (chicken, pork, and beef), 126 raw milk samples from raw milk cheese manufacturers, and 135 ready-to-eat foods (meat products, salads, and raw milk cheeses). Campylobacter jejuni was detected in 4 of the 316 samples analyzed: 1 raw beef liver sample and 3 raw chicken samples. An isolation rate of 9.7% was observed among the raw chicken samples tested. This study also investigated the role of cross-contamination in disseminating Campylobacter from raw poultry within a food service operation specializing in poultry dishes. Accordingly, kitchen surfaces within a restaurant in Ottawa, Ontario, were sampled between March and August 2001. Tests of the sampling method indicated that as few as 100 Campylobacter cells could be detected if sampling was done within 45 min of inoculation; however, Campylobacter spp. were not detected in 125 swabs of surfaces within the kitchens of this food service operation. Despite the reported high prevalence of Campylobacter spp. in raw poultry, this organism was not detected on surfaces within a kitchen of a restaurant specializing in poultry dishes.     

AQUA: An Amphibious Autonomous Robot

AQUA, an amphibious robot that swims via the motion of its legs rather than using thrusters and control surfaces for propulsion, can walk along the shore, swim along the surface in open water, or walk on the bottom of the ocean. The vehicle uses a variety of sensors to estimate its position with respect to local visual features and provide a global frame of reference     

Palladium‐Catalyzed C?H Functionalization of Ferrocenecarboxylic Acid by using 8‐Aminoquinoline as a Removable Directing Group

A mild and efficient palladium‐catalyzed synthetic method for the C H functionalization of N‐(quinolin‐8‐yl)ferrocenecarboxamide has been developed. Various aryl iodides containing I, NO₂, CN, COMe, CO₂Et, and NH functionalities and also alkyl iodides underwent the Pd‐catalyzed intermolecular carbon‐carbon bond forming reaction with ferrocenecarboxamide successfully which led to a diverse array of bis(aryl/alkyl)ferrocenecarboxamides in 34–92% yields. Cross‐coupling of the ferrocenyl C H bond with aryl iodides can also be achieved utilizing an economical Ni catalyst. Additionally, selective monoalkylation of ferrocenecarboxamide was studied using sodium bicarbonate as base and dibenzylphosphoric acid as additive under Pd‐catalyzed reaction conditions. Subsequently, removal of the directing group, 8‐aminoquinoline, from bis(aryl)ferrocenecarboxamides led to bis(aryl)ferrocenes bearing versatile methyl ester and carboxaldehyde functional groups.