※日本語版が続きます。
About AIRoA
AI Robot Association (AIRoA) is an organization dedicated to collecting large-scale real-world robot data, including data from humanoid robots, and advancing the development of generative AI foundation models for robotics.
AIRoA has been selected as a project operator for the development of a data platform for generative AI foundation models in robotics under the “Post-5G Information and Communication Systems Infrastructure Enhancement R&D Project” by Japan’s Ministry of Economy, Trade and Industry (METI) and NEDO. The project budget is 20.5 billion yen. Based on this platform, AIRoA is carrying out a project to collect approximately one million hours of humanoid robot operation data using more than 100 robots, and to develop a world-class Vision-Language-Action (VLA) model using that data.
The significance of this initiative lies not only in the scale of its real-world data and humanoid platforms. AIRoA aims to make datasets, trained models, benchmarks, and evaluation environments as open as possible, building a “Robot Data Ecosystem” that can be commonly used by researchers, startups, and large enterprises. Achieving this requires not only AI models and cloud infrastructure, but also the capability to operate physical robots reliably and integrate sensors, actuators, control systems, and communication systems with a high degree of reliability through embedded software and firmware.
In this position, you will be responsible for firmware and embedded software development for humanoids, dual-arm robots, robotic hands, teleoperation and leader–follower systems, and robot systems for data collection used in AIRoA’s research projects.
Responsibilities
-
Simulation infrastructure: Design and implement simulation environments, scenarios, evaluation tools, and log replay functions for humanoids, mobile robots, and service robots.
-
Automated testing and CI: Establish simulation-based software testing, automated regression testing, and CI/CD integration to detect performance degradation and safety issues caused by software changes at an early stage.
-
Evaluation scenarios: Build a scenario library and benchmark suite that reproduce real-world tasks, failure cases, sensor abnormalities, contact, crowded environments, robot state transitions, and related situations.
-
Synthetic data and sensor simulation: Prepare synthetic data, sensor simulation, domain randomization, and log-derived scenarios for use in perception, navigation, and VLA evaluation.
-
Sim-to-real analysis: Measure differences between physical robot logs and simulation results, and improve models, sensors, control systems, environment representations, and evaluation metrics.
AIRoAについて
AI Robot Association（AIRoA）は、ヒューマノイドロボットをはじめとする実世界ロボットの⼤規模データを収集し、ロボティクス分野における⽣成AI基盤モデルの開発を推進する組織です。
AIRoAは、経済産業省およびNEDOの「ポスト5G情報通信システム基盤強化研究開発事業」において、ロボティクス分野の⽣成AI基盤モデル開発に向けたデータプラットフォーム開発の採択事業者に決定しており、事業予算は205億円です。この基盤をもとに、100台以上のロボットを⽤いて100万時間規模のヒューマノイドロボット操作データを収集し、それを活⽤して世界最⾼⽔準のVision-Language-Action（VLA）モデルを開発するプロジェクトを進めています。
この取り組みの特徴は、実世界データとヒューマノイドプラットフォームの規模だけではありません。AIRoAは、データセット、学習済みモデル、ベンチマーク、評価環境を可能な限りオープンにし、研究者‧スタートアップ‧⼤企業が共通して活⽤できる「Robot Data Ecosystem」の構築を⽬指しています。その実現には、AIモデルやクラウド基盤だけでなく、実機ロボットを安定して動作させ、センサー‧アクチュエータ‧制御系‧通信系を⾼い信頼性で統合する組み込みソフトウェア∕ファームウェアの⼒が不可⽋です。
本ポジションでは、AIRoA内の研究プロジェクトで使⽤するヒューマノイド、双腕ロボット、ロボットハンド、遠隔操作‧leader–follower system、データ収集⽤ロボットシステムにおける、ファームウェアおよび組み込みソフトウェア開発を担っていただきます。
業務内容
-
シミュレーション基盤: ヒューマノイド、モバイルロボット、サービスロボット向けのシミュレーション環境、シナリオ、評価ツール、ログリプレイ機能を設計‧実装します。
-
⾃動テスト‧CI: simulation-based software testing、automated regression testing、CI/CD連携を整備し、ソフトウェア変更による性能劣化や安全上の問題を早期に検知します。
-
評価シナリオ: 実環境タスク、失敗ケース、センサ異常、接触、混雑環境、ロボット状態遷移などを再現するscenario libraryとbenchmark suiteを構築します。
-
合成データ‧sensor simulation: perception、navigation、VLA評価に利⽤する、合成データ、sensor simulation、domain randomization、ログ由来シナリオを整備します。
-
sim-to-real分析: 実機ログとシミュレーション結果の差分を測定し、モデル、センサ、制御、環境表現、評価指標を改善します。
Requirements
Required Qualifications
-
Experience developing or operating robotics, autonomous driving, factory automation, game engine, physics simulation, or high-fidelity simulation environments.
-
Professional experience using simulators such as Isaac Sim, MuJoCo, Gazebo, PyBullet, Unity, or Unreal Engine.
-
Experience using Python for evaluation pipelines, data processing, simulation tools, log analysis, visualization, and test automation.
-
Experience designing mechanisms that development teams can continuously use, such as test cases, evaluation metrics, log replay, benchmarks, CI, and automated reports.
-
Ability to understand one or more of the following areas and translate them into simulation requirements: sensors, control, state transitions, navigation, manipulation, or real-world robot constraints.
Preferred Qualifications
-
Ability to organize requirements across multiple teams such as Autonomy, VLA, Navigation, Integration, and Hardware, and implement simulation environments as a shared platform.
-
Experience with cloud-based simulation infrastructure, GPU-accelerated simulation, containerization, distributed execution, and job management.
-
Experience with imitation learning, reinforcement learning, ML policy training, VLA evaluation, or synthetic data pipelines.
-
Experience with ROS/ROS2, bag/log replay, Nav2, MoveIt, robot descriptions, sensor plugins, or simulation-to-real integration.
-
Experience with Gaussian splatting, NeRF, world models, learned sensor models, diffusion-based methods, or neural rendering.
必須要件
-
ロボティクス、⾃動運転、FA、ゲームエンジン、物理シミュレーション、または⾼忠実度シミュレーション環境の開発‧運⽤経験があること。
-
Isaac Sim、MuJoCo、Gazebo、PyBullet、Unity、Unreal Engine等を⽤いたシミュレータを実務で使⽤した経験
-
Pythonを⽤いた評価パイプライン、データ処理、シミュレーションツール、ログ解析、可視化、テスト⾃動化の経験があること。
-
テストケース、評価指標、ログリプレイ、benchmark、CI、⾃動レポートなど、開発チームが継続利⽤する仕組みを設計した経験があること。
-
センサ、制御、状態遷移、ナビゲーション、マニピュレーション、実機制約のいずれかを理解し、シミュレーション要件に落とし込めること。
歓迎要件
-
Autonomy、VLA、Navigation、Integration、Hardwareなど複数チームの要件を整理し、シミュレーション環境を共通基盤として実装できること。
-
cloud-based simulation infrastructure、GPU-accelerated simulation、containerization、分散実⾏、ジョブ管理の経験。
-
imitation learning、reinforcement learning、ML policy training、VLA評価、synthetic data pipelineの経験。
-
ROS/ROS2、bag/log replay、Nav2、MoveIt、robot description、sensor plugin、simulation-to-real integrationの経験。
-
Gaussian splatting、NeRF、world models、learned sensor models、diffusion-based methods、neural renderingの経験。
Benefits
There are currently no comparable projects in the world that collect data and develop foundation models on such a large scale. As mentioned above, this is one of Japan’s leading national projects, supported by a substantial investment of 20.5 billion yen from NEDO.
This position will play a crucial role in determining the success of the project. You will have broad discretion and responsibility, and we are confident that, if successful, you will gain both a great sense of achievement and the opportunity to make a meaningful contribution to society.
Furthermore, we strongly encourage engineers to actively build their careers through this project—for example, by publishing research papers and engaging in academic activities.
●Work location
Tokyo Ryutsu Center A Bldg. AW4-5, 6-1-1 Heiwajima, Ota-ku, Tokyo 143-0006, Japan
