M
MERL is Mitsubishi Electric's US R&D subsidiary, focused on AI, optimization, and control theory applied to physical systems. The tech stack—PyTorch, JAX, MuJoCo, ROS 2, MATLAB, Gurobi—reveals a heavy emphasis on physics simulation and robotics alongside deep learning, with active projects spanning safe reinforcement learning, vision-language models, and industrial anomaly detection. The organization is research-heavy (53 research roles vs. 12 engineering) with heavy intern recruitment, typical of an academic-adjacent lab that treats hiring as pipeline-building.
Tech Stack 38 technologies
Core StackPython PyTorch C++ TensorFlow MATLAB JAX DETR nuScenes ROS 2 C/C++ CLIP PyBullet MuJoCo ROS CasADi Slurm dSPACE Julia Modelica OpenVLA Mujoco Isaac Sim GTSAM Gurobi Gazebo PX4 Simulink OpenFOAM pybind11 Simscape+7 more
What Mitsubishi Electric Research Laboratories Is Building
◆Challenges
- Detecting network failure in multi-hop iot networks
- Improving efficiency and scalability of carbon capture
- Evaluating novel carbon capture concepts
- Efficient implementation of data assimilation algorithms
- Privacy and security of agentic ai systems
▲Active Projects
- Design experiments with in-house testbed
- Sample efficient safe reinforcement learning for ultra-high precision positioning systems
- State estimation
- Hybrid vehicle analysis and optimization
- Novel vision + language model architectures
- Generative audio research
- Diffusion and flow matching-based architectures
- Anomaly localization in industrial inspection
- Develop algorithms for large foundation models in image anomaly localization
- Bos tomography 3d airflow reconstruction
Hiring Activity
Minimal65 roles · 7 in 30d
Department
Seniority
Company intelligence