Agri Embodied AI を考える時


そこで、日本農林資源開発では、ノウハウを持たない新規参入者が一定の品質の作物を生産できるようにするための農業AIアシスタントシステム「Agri Embodied AI Edge Data Center」を提案します。


Agri Embodied AI エッジデータセンターは、圃場のセンサー情報を基に機械学習を行うと共に農業従事者とこの地方特有の栽培方法など自然言語で対話し生成AIによりインタラクティブな予測値の可視化を提供します。これにより、環境に基づいた対話と学習を通じて、農業作業のナビゲーションを支援し、自己の行動の長期的な結果を考慮して作業を計画し、それを可視化します。



Currently, agriculture is advancing digitization through the use of smart agricultural machinery, sensor-based field visualization and management, and satellite-based field change prediction. Improving food self-sufficiency is crucial for our country amidst rising nationalism worldwide and stricter food import conditions. However, we also face challenges such as aging agricultural workers and declining workforce numbers.

To address these issues, Japan Agriculture and Forestry Resource Development Co,. propose the “Agri Embodied AI Edge Data Center,” an agricultural AI assistant system aimed at enabling new entrants without specialized knowledge to produce crops of consistent quality.

The ultimate goal of this advanced agricultural AI assistant is to develop agents capable of creatively solving challenging tasks that require interaction with the environment. This is a significant challenge, but advances in deep learning and the availability of large-scale datasets may allow for superhuman performance in agricultural tasks that were previously considered insurmountable.

The Agri Embodied AI Edge Data Center leverages machine learning based on field sensor data and facilitates interactive visualization of predictive values through generated AI, enabling natural language conversations with farmers about region-specific cultivation methods. This supports navigation through agricultural tasks by integrating environmental dialogue and learning, allowing for long-term planning and visualization of one’s actions and their outcomes.

The instantiated AI agent guidance unit must be capable of adapting to unforeseen situations and problem-solving within trained tasks and goal achievement.

This is a big goal, but we believe that once we get started, we can put it into practical use within about five years.



No responses yet


    メールアドレスが公開されることはありません。 が付いている欄は必須項目です

    Latest Comments