PhD - Neural 3D Scene Representations for the Automotive Cabin Location: Hildesheim, Lower Saxo[...]

Boschpackaging; Hildesheim; Vor Ort; EUR 70.000 - 95.000; Vollzeit

Zusammenfassung

Robert Bosch GmbH's Research unit is seeking a motivated researcher to advance 3D perception for in-cabin and vehicle environments. You will explore neural 3D scene representations and leverage cabin geometry to supervise learning, while integrating 3D priors and diffusion models.

We target reconstructing occupants from limited camera viewpoints to support end-to-end ADAS systems and safety-critical applications, collaborating with academia and industry. Start: October 2026.

Qualifikationen

• Experience in machine learning using PyTorch, ideally with vision-based models and neural scene representations.
• Programming in Python; solid foundations in computer vision and 3D geometry.
• Familiarity with 3D scene representations (e.g., NeRF, Gaussian splatting, implicit/occupancy networks) is preferred.

Aufgaben

• Investigate neural 3D scene representations and adapt them to in-cabin perception challenges.
• Leverage cabin geometry as supervision and incorporate 3D-aware diffusion priors.
• Research reliable reconstruction of occupants from one to three camera viewpoints for end-to-end ADAS systems.
• Collaborate with Bosch Research and academia to advance safety-related products.

Kenntnisse

PyTorch

Python

3D geometry

NeRF/3D scene representations

65%

Ausbildung

Master’s degree in Computer Science

Jobbeschreibung

At Bosch, we shape the future by inventing high-quality technologies and services that spark enthusiasm and enrich people's lives. Our promise to our associates is rock-solid: We grow together, we enjoy our work, and we inspire each other. Welcome to Bosch.

The Robert Bosch GmbH is looking forward to your application!

At Bosch Research, we are driving the next generation of Physical AI, intelligent systems capable of perceiving, reasoning about, and acting robustly in real-world environments. As vehicles become increasingly shared, automated, and software-defined, understanding what happens inside the cabin is becoming as important as understanding the road ahead. The ability to detect occupant position, posture, and movement is evolving into both a regulatory requirement and a key enabler of future safety and comfort functions. However, current perception systems largely treat the vehicle interior and exterior as separate domains, limiting their ability to form a coherent three-dimensional understanding of the cabin from the small number of cameras that can realistically be deployed in a vehicle.

Responsibilities

• Investigate neural 3D scene representations, including neural radiance fields, Gaussian splatting, and implicit occupancy models, and adapt them to the unique challenges of in-cabin perception.
• Leverage the asymmetry of the cabin environment by using the known cabin geometry as built-in supervision and by incorporating modern 3D-aware diffusion models as learned priors.
• Research the reliable reconstruction of occupants from as few as one to three fixed camera viewpoints, and create a unified representation of the vehicle interior and exterior to enable end-to-end ADAS systems that reason about passengers and surrounding traffic.
• Work at the forefront of AI innovation and contribute to real-world industrial applications within Bosch Research.
• Collaborate closely with leading experts from academia and industry to bring advances into future safety‑critical products.
• Combine fundamental research in 3D computer vision and generative modeling with the practical goal of making vehicles safer for everyone on board.

Qualifications

• Education: Excellent Master’s degree in Computer Science or a comparable field of study.
• Experience and Knowledge:
  • Experience in machine learning using PyTorch, ideally with vision-based models and neural scene representation.
  • Programming skills in Python and solid foundations in computer vision and 3D geometry, including camera models, projective geometry, and multi-view constraints.
  • Familiarity with 3D scene representations (e.g., NeRF, Gaussian splatting, implicit/occupancy networks) or depth/3D estimation methods is strongly preferred.
  • Prior exposure to differentiable rendering or articulated human pose estimation is a plus.
• Personality and Working Practice: Proactive and highly motivated mindset with the ability to confidently navigate complex technical environments.
• Languages: Good English skills, both written and spoken.

The final Ph.D. topic is subject to the supervising university.

Start: October 2026

Diversity and inclusion are not just trends for us but are firmly anchored in our corporate culture. Therefore, we welcome all applications, regardless of gender, age, disability, religion, ethnic origin or sexual identity.