Computer Vision Team

At a Glance

The Computer Vision Team The Computer Vision team develops 3D perception, object detection, and camera calibration algorithms to give robots sight.

Team Mission

We build high-performance automation software and systems at Mujin, enabling autonomous perception, motion planning, and real-time execution.

Core Responsibilities

We build the eyes and perceptual logic of Mujin’s automated solutions:

3D Pose Estimation

We develop fast and precise algorithms to detect and estimate the exact position and orientation of complex items in 3D space.

Camera Calibration

We build automated, highly accurate calibration methods to align 3D cameras and sensors with the robot’s physical coordinate system.

Deep Learning Deployment

We train and optimize deep learning models, deploying them on edge GPUs to handle segmentation and classification in milliseconds.

---

Tech Stack & Workflows

We design latency-critical vision pipelines that run on GPUs:

1. Programming Languages & Frameworks
   • C++ and CUDA for real-time 3D processing pipelines.
   • Python, OpenCV, and PyTorch for deep learning development and data tooling.
2. Sensors & Hardware Acceleration
   • Industrial 3D sensors, RGB-D cameras, and stereo vision systems.
   • Nvidia TensorRT for high-speed model inference at the edge.
3. Development & Validation Workflow
   • Testing algorithms on datasets compiled from diverse factory environments.
   • Physical test cells in our office to evaluate vision accuracy in real-time.

Team Fact

Our vision pipelines can process high-resolution 3D point clouds and return coordinates in under 200 milliseconds, matching the fastest robot cycle times.

---

Who Succeeds in Our Team

We look for engineers with strong analytical and geometric foundations:

• Technical Competence: Strong command of 3D geometry, linear algebra, and high-performance algorithms.
• Effective Collaboration: Working closely with Robotics teams to translate vision outputs into smooth trajectories.
• Continuous Learning: Reading the latest CV papers and adapting state-of-the-art models for industrial use.

---

Open Positions

If you want to solve complex real-world perception challenges, apply today:

View Open PositionsCheck out our active job openings in Tokyo and become part of our international team.