Robotic Perception Expert - Arms (human)
Neura Robotics · Metzingen, Riederich
Give Our Arms Sight and Touch
Perception is where a robot's intelligence starts. In our AI/Perception team, you'll own the algorithms that turn raw camera, depth, and force data into something an arm can act on — detecting and tracking objects, estimating 6D pose, reconstructing the workspace, and fusing it all into a picture reliable enough to close a gripper on. Our arms are built to work alongside people in workspaces that never hold still, so this isn't benchmark work: your models run on real hardware, next to real humans, and have to hold up when the lighting is poor, the bin is cluttered, and the scene changes mid-task. You'll shape the perception stack from architecture to deployment, working closely with our hardware, control, planning, and manipulation teams. If you want to spend your time on perception problems that don't have clean answers yet, this is the place.
Your Mission & Challenges
We are looking for an experienced Robotic Perception Expert to join our AI/Perception team. You will architect and build the perception stack that lets our collaborative robot arms see, understand, and interact with dynamic, shared workspaces — and you will see it run on products, not just in papers.
You will take ownership of areas such as:
Perception pipeline: Architect and build the full perception path — from raw sensor input (RGB and depth cameras, force/torque, joint sensors) through sensor fusion, object detection and tracking, 6D pose estimation, 3D reconstruction, and scene understanding — to produce the workspace models our arms rely on for grasp planning, manipulation, and safe interaction with the people working next to them.
Research to product: Track the state of the art in perception and manipulation, and bring what works into real robot systems running in real cells.
Deep learning for perception: Design, train, and deploy neural networks for perception tasks, and get them running within the latency and compute budget of embedded robotic hardware.
Computer vision & 3D: Build and optimize computer vision algorithms for real-time use, including classical vision (OpenCV), 3D geometry, hand-eye calibration, and point-cloud processing (PCL).
Real-time software: Implement perception software in modern C++ and Python, using GPU acceleration where it matters.
ROS2 & DDS: Build high-performance sensor pipelines with safety and robustness as first-class concerns.
Cross-team collaboration: Work hand in hand with the hardware, control, planning, and manipulation teams to turn perception output into motion.
What We Can Look Forward To
Education: Excellent Master's or PhD in Computer Science, Robotics, Electrical Engineering, or a related field
Experience: 3+ years of relevant experience building robust perception stacks for human-centered environments
Programming skills: Strong programming skills in Python and modern C++; CUDA is a plus
Deep learning: Hands-on experience with deep learning frameworks (PyTorch, TensorFlow)
Computer vision background: Solid background in computer vision and sensor fusion: 3D geometry, RGB-D and force-torque sensing, PCL, OpenCV
Robotics software: Practical experience with ROS2 and real-time robotic software
Manipulation perception: Experience with manipulation-oriented perception — grasp detection, pose estimation, bin picking, or workspace monitoring — is a strong plus
Mindset: Strong problem-solving and analytical skills
Communication: Your communication skills make you shine
Language: You have a perfect command of the English language