Evaluation of multiple satellite precipitation products for rainfed maize production systems over Vietnam

Paresh Shirsath Dakshina Murthy Kadiyala (2022, [Artículo])

Rainfall Datasets Satellite Rainfall Estimates CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA RAIN RAINFED FARMING DATA SATELLITES

On-farm storage loss estimates of maize in Kenya using community survey methods

Hugo De Groote Anani Bruce (2023, [Artículo])

Maize is the most important staple in sub-Saharan Africa (SSA), with highly seasonal production. High storage losses affect food security, but good estimations are lacking. A new method using focus group discussions (FGDs) was tested with 121 communities (1439 farmers, 52% women) in Kenya's six maize-growing zones, to estimate the maize losses to storage pests and analyze farmer practices. As control strategies, half of the farmers used chemical pesticides (49%), while hermetic bags (16%) and botanicals (15%) were also popular. Relative loss from weevils in the long rains was estimated at 23%, in the short rains 18%, and annually 21%. Fewer farmers were affected by the larger grain borer (LGB) than by maize weevils: 42% in the long rainy season and 32% in the short rainy season; losses from LGB were also smaller: 19% in the long season, 17% in the short season, and 18% over the year. Total storage loss, from both species combined, was estimated at 36%, or 671,000 tonnes per year. The greatest losses occur in the humid areas, especially the moist mid-altitudes (56%), and with smaller loss in the drylands (20–23%). Extrapolating the point data and overlaying with the maize production map shows the geographic distribution of the losses, with the most important area found around Lake Victoria. FGDs provide convenient and cheap tools to estimate storage losses in representative communities, but a total loss estimate of 36% is higher than is found in other studies, so its accuracy and framing effects need to be assessed. We conclude that storage pests remain a major problem, especially in western Kenya, and that the use of environmentally friendly technologies such as hermetic storage and botanicals needs more attention, both by the public extension service and private agrodealers.

Larger Grain Borer Maize Weevil CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA MAIZE STORAGE LOSSES PESTS SURVEY METHODS

Novel mutant alleles reveal a role of the extra-large G protein in rice grain filling, panicle architecture, plant growth, and disease resistance

Akshaya Biswal Daisuke Urano (2022, [Artículo])

Heterotrimeric G Proteins Extra-Large G Proteins Cas9 OsXLG CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA CRISPR RICE PROTEINS PLANT GROWTH DISEASE RESISTANCE

Detección de comportamiento no verbal en interacción humano-robot

Detection of non-verbal behavior in human-robot interaction

Ernesto Adrián Lozano De la Parra (2023, [Tesis de maestría])

La comunicación no verbal desempeña un papel vital en la interacción humana. En el contexto de la interacción humano-robot (IHR), los robots sociales están diseñados principalmente para la comunicación verbal con los humanos, dejando a la comunicación no verbal como un área de investigación abierta. En este trabajo, se presenta una arquitectura flexible y abierta llamada Software Arquitechture for Nonverbal Interaction in Human-Robot Interaction (SANI-HRI) diseñada para facilitar las interacciones no verbales en IHR. Entre sus componentes se encuentra un Cuaderno Computacional P2P basado en navegador web, aprovechado para codificar, ejecutar y compartir programas reactivos. Pueden incluirse modelos de aprendizaje automático para el reconocimiento en tiempo real de gestos, poses y estados de ´animo, empleando protocolos como MQTT. Otro componente clave es un Broker para distribuir datos entre distintos dispositivos físicos, como robots, dispositivos vestibles y sensores ambientales, así como modelos de aprendizaje automático que comprendan diferentes tipos de datos. Se demuestra la utilidad de esta arquitectura mediante tres escenarios de interacción: (i) el primero que emplea la proxémica y la dirección de la mirada para iniciar un encuentro improvisado, (ii) un segundo que utiliza técnicas de visión por computadora para detectar y analizar expresiones faciales y corporales, así como el uso sensores biométricos para obtener datos de ritmo cardiaco durante una rutina de ejercicio, y (iii) un tercero que incorpora el reconocimiento de objetos y Modelos de Lenguaje Grandes para sugerir comidas a cocinar en función de los ingredientes disponibles. Estos escenarios ilustran cómo los componentes de la arquitectura pueden integrarse para abordar nuevos escenarios, en los que los robots necesitan inferir señales no verbales de los usuarios.

Nonverbal communication plays a vital role in human interaction. In the context of Human-Robot Interaction (HRI), social robots are designed primarily for verbal-based communication with humans, making nonverbal communication an open research area. We present a flexible, open framework called Software Architecture for Nonverbal Interaction in Human-Robot Interaction (SANI-HRI) designed to facilitate nonverbal interactions in HRI. Among its components it has a P2P Browser-Based Computational Notebook, leveraged to code, run, and share reactive programs. Machine-learning models can be included for real-time recognition of gestures, poses, and moods, employing protocols such as MQTT. Another key component is a broker for distributing data among different physical devices like the robot, wearables, and environmental sensors and also machine learning models. We demonstrate this framework’s utility through three interaction scenarios: (i) the first one employing proxemics and gaze direction to initiate an impromptu encounter, (ii) a second that uses computer vision techniques to detect and analyze facial and body expressions, as well as the use of biometric sensors to obtain heart rate data during a workout routine, and (iii) a third one incorporating object recognition and a Large-Language Model to suggest meals to be cooked based on available ingredients. These scenarios illustrate how the framework’s components can be seamlessly integrated to address new scenarios, where robots need to infer nonverbal cues from users.

Interacción humano-robot, Comunicación no verbal, Broker MQTT, Notebook computacional, Modelos linguísticos grandes, SANI-HRI Human-robot interaction, Nonverbal communication, Broker MQTT, Computational notebook, Large language models, SANI-HRI INGENIERÍA Y TECNOLOGÍA CIENCIAS TECNOLÓGICAS TECNOLOGÍA DE LOS ORDENADORES SISTEMAS DE RECONOCIMIENTO DE CARACTERES SISTEMAS DE RECONOCIMIENTO DE CARACTERES