Several factors render this combination challenging to achieve. Living tissues can be easily damaged and handling them typically requires slow, careful manipulation that avoids excessive forces or pressures. Biological variation introduces large variability in physical properties such as shape, size, mass, firmness of the targeted plants or plant components. This variability, coupled with uncertainty in the sensing system and limitations in the performance of control systems can affect negatively the accuracy, speed, success rate and effectiveness of the operation. Reduced accuracy can cause damage to the targeted part of the plant or nearby plant parts , or the entire plant . It may also cause reduced throughput due to misses and repeats, or reduced efficiency if no repeats are attempted. Visual servoing/guidance of robot actuators can reduce uncertainty and increase efficiency, but uneven illumination, shadows cast by branches and leaves, partial occlusions, and branches acting as obstacles present significant challenges in real-world conditions . Guiding the end-effector by combining inputs from multiple cameras is an approach that could be adapted to agricultural settings . Another possible direction is using deep reinforcement learning to learn visual servoing that is robust to visual variation, changes in viewing angle and appearance, and occlusions .If stem-cutting is used, challenges include detecting and cutting quickly and robustly from a large range of approaches, in the presence of touching fruits and twigs. If pulling is used, the force required to detach fruits depends on the type and maturity of the fruits, the approach angle of the end-effector, and on whether rotation is also used. Some fruits require concurrent, controlled, vertical grow rack synchronized rotation and pulling to reduce skin/peel damage at the stem-fruit interface , a task that is complex and not easily modeled. Deep reinforcement learning for grasping is a possible approach to build sophisticated controllers for such tasks.
Innovations in materials, design and control for soft robots could also be adapted to fruit picking and crop handling in general . Another important factor is limited accessibility of the targeted plants or their parts by robot end-effectors. Accessibility can be limited by plant structure, positioning, interference with neighboring plants or structures, and robot design. For example, in robotic weeding, weeds that are very close to a crop-plant’s stem and hidden under its canopy are not easily accessible by the end-effector without damaging the crop . In fruit harvesting, fruits in tree canopies that are positioned behind other fruits, branches or trellis wires also have limited accessibility by robotic harvesting arms. Accessibility can be improved by introducing dexterous, multi-d of actuation systems. However, control complexity can reduce throughput; the overall system cost will also be higher. Breeding and horticultural practices can also be utilized to improve accessibility. For example, tree cultivars with smaller and simpler canopies, training systems that impose simpler – planar – canopy geometrical structures along fruit thinning operations can contribute to higher fruit accessibility/reach ability. To some extent, it is the availability of trellised planar architectures and precision fruit thinning which result in very high fruit visibility and reach ability that have enabled robotic harvesting to emerge recently as a potentially cost effective approach to mechanical fruit harvesting at commercial scale. However, the cost and required labor demand for maintaining meticulously thinned and pruned trellised trees can be very high. Moreover, not all fruit trees can be trained in such narrow, planar systems. A promising approach that can be used to guide “breeding for manipulation” is the use of plant and robot geometric models to co-design tree structures and machines to optimize manipulation reach ability and throughput . Also, the use of large numbers of simpler, cheaper actuators that approach plants from different positions has shown promise in terms of reach ability , and could be adopted to increase overall throughput.Endotoxins are lipopoly saccharides in the outer membranes of Gram-negative bacteria that are distributed widely on plants, in soil, water, and the intestines of humans and animals [reviewed by Myatt and Milton ; Spaan et al. ].
Endotoxins are found in indoor dust generated by human activity and pets and are also found adsorbed onto the surfaces of combustion particles Inhaled endotoxins are bound by an LPSbinding protein that, in turn, binds to specific cell receptor [CD14 , a Toll-like receptor ], and initiates signaling pathways that lead to expression of proinflammatory cytokines that result in lung inflammation, increases in epithelial permeability, and activation of systemic inflammation . Although high concentrations of aerosolized endotoxin have been recognized as a cause of lung disease in cotton workers and swine handlers , recent interest has focused on the complex role of nonoccupational indoor and outdoor endotoxin concentrations in the occurrence of immunoglobulin E –mediated allergy and asthma . Biological responses to endotoxin, in theory, could lead both to suppression of IgE-mediated responses through the stimulation of interleukin 12 and to the worsening of airway inflammation, a hallmark of asthma . These effects have been reported at endotoxin concentrations lower than those found in high-risk occupational settings. Several studies have associated elevated levels of house dust endotoxins with a) increased respiratory symptoms in infants ; b) worsening of existing asthma that is independent of the levels of other common indoor allergens ; c) decreased frequency of positive IgEmediated skin test reactions in infants ; and d) decreased occurrence of hay fever and positive prick skin test in children . Rural residence, particularly on farms with animal exposure, has been reported to reduce risk of asthma . Despite the known high levels of endotoxin in these settings , definitive evidence that endotoxin, and not some other component of the microbial flora, is associated with this decreased risk has not been established . Most studies of the association between human exposure to endotoxins and allergic and respiratory disease have focused on concentrations of endotoxin in samples of house dust . Few studies have evaluated the correlation between endotoxin concentrations in dust and air , which appears to be low— correlation < 0.3 . Several recent studies have described ambient concentrations of endotoxin. Endotoxin concentrations in New Orleans, Louisiana, after flooding from Hurricane Katrina were high in flooded [3.9 EU /m3] and non-flooded areas and did not differ between indoor and outdoor environments . Ambient endotoxin concentrations in a large area of Southern California were below a 5.5-EU/m3 limit for adverse health effects in occupational settings quoted by the authors . The highest endotoxin content per milligram of PM10 was found in the mountain and desert areas. No seasonal patterns were detected. A 5.5-month study at the University of North Carolina found that ambient endotoxin concentrations were greater in coarse particles [aerodynamic diameters between 2.5 and 10 µm ] than in particles with aerodynamic diameters < 2.5 µm . An extensive study of size-fractionated bioaerosol was performed in 20 homes in and around Palo Alto, California . During the daytime, the highest concentrations of endotoxin were in particles with aerodynamic diameters > 10 µm , followed by the PMc size fraction. At night, the highest concentrations occurred in the PMc size fraction. Of the above studies, only the study in Southern California provides some data on spatial distributions of endotoxin based on where subjects resided; however,grow light racks potential ambient sources were not investigated . As part of a study of the effects of ambient air pollution on the natural history of children with asthma, we characterized the temporal and spatial distributions of ambient endotoxin over several years in Fresno and Clovis, California , a city surrounded by large tracts of land devoted to agriculture and animal husbandry. As part of a study to evaluate the role of ambient air pollution and bioaerosols on the natural history of childhood asthma, in this article, we focus on ambient endotoxin, its spatial distribution in relation to these sources, and the influence of meteorologic factors on daily concentrations. Fresno is located in the San Joaquin Valley near the southern end of the Central Valley of California. In 2006, the population was 466,700. The study area was confined to a circle with a radius of 20 km, with its center at the ambient air monitoring station operated by the California Air Resources Board .
The city is bound on three sides by land used primarily for agriculture and in the northeast by native vegetation. Two major interstate highways cross the study area: California State Highway 99 from northwest to southeast and Interstate 41 from north to south. The wind patterns are variable [see Supplemental Material ]. For data on collection of ambient concentration, see Supplemental Material, Figure S1 .Daily ambient endotoxin was collected year-round at the California ARB central ambient monitoring site at 3425 First Street in Fresno as part of the exposure assessment for the Fresno Asthmatic Children’s Environment Study . FACES is a cohort of 315 children 6–11 years of age at enrollment with clinically active asthma. All subjects lived within a 20-km radius of a U.S. Environmental Protection Agency Super Site located in Fresno. Subjects were followed with biannual evaluations of respiratory health, pre- and post bronchodilator spirometry, skin prick testing and household surveys. Subjects also completed three 14-day panel studies over three seasons based on ambient pollution concentrations in the study area. Initially, the samples were collected at the First Street site from midnight to midnight. In early 2002, the collection times were changed to 2000 to 2000 hours to coincide with the times that data were collected during panel studies and the times of the intensive sampling of 83 homes selected to cover the full range of indoor and outdoor exposures in the study community. Daily samples reported here cover 13 May 2001 through 31 October 2004. Additional samples were collected from June 2002 to August 2003 at 10 school locations , with two mobile trailers outfitted by the ARB to include the instrumentation identical to that located at the First Street site. In parallel, ambient endotoxin samples were collected inside and outside 83 homes between 6 February 2002 and 22 February 2003 over 5 days during the 2-week panel studies of the children. Twenty eight homes were sampled twice during two separate panels in two seasons . Concentrations were also measured at each location for elemental carbon , PM2.5, and PM10. Concentrations of PMc were determined by the difference . On the residential sampling days, 24-hr samples were collected at up to eight locations: First Street, Fremont School, one other school, and up to five residences . At First Street and the schools, airborne endotoxin was collected on 47-mm Teflon filters in a Partisol-Plus Model 2025 Sequential Air Sampler with a PM10 inlet . Samples were collected at a nominal flow rate of 8.33 L/min for 24 hr. At residences, 24-hr integrated samples were collected with Harvard type PM10 impactors at 10 L/min flow rate in a multileg sampler. One sampling leg used 37-mm Teflon filters for determination of PM10 mass and endotoxins. The other sampling legs employed inlets and filter media for determination of PM2.5 mass, sulfate and nitrate, organochlorines and EC, nicotine, metals, and polycyclic aromatic hydrocarbons. Filters were loaded and unloaded in 24-hr periods before and after the sampling period and sent to the laboratory for analysis. Collocated endotoxin data collected with the two different samplers differed by < 0.1 EU/m3, on average .We confined our analysis to 45 of 107 residential sampling days that also had First Street endotoxin measurements and occurred during the dry season for the reasons stated previously. However, for spatial mapping of concentration patterns, we further restricted our analysis to days when six or more locations from all sites had data, which reduced the data set to 22 dry-season sampling days. The sparseness of the data at most sampling locations limits the applications of conventional spatial analysis methods; nonetheless, the data are sufficient to describe a) the relations between concentrations at schools and the central air monitoring station using regression equations and coefficients of divergence [see Supplemental Material, Equation 1 ]; b) the range of daily spatial variability across the urban area using coefficients of spatial variations ; and c) the average spatial patterns.