PhD - Predictive Occupancy World Models for End-to-End Autonomous Driving

As PhD candidate in Bosch Research, you will advance AI-driven mobility by building predictive occupancy models that capture scene geometry and semantics for robust 4D world understanding. You will design and test end-to-end driving architectures that integrate perception with driving decisions, using simulation and real-world data. You will contribute to safe, transparent autonomous systems and publish findings at top AI/robotics venues. This role blends fundamental research with practical evaluation to push the boundaries of autonomous driving.

• Develop a generic predictive occupancy model as the core representation of driving scenes
• Design architectures that capture geometry and semantics for 4D world dynamics
• Research, implement, and evaluate end-to-end driving architectures based on predictive occupancy models
• Develop perception systems that convert raw sensor data into spatial representations and driving actions
• Create architectures where driving decisions hinge on predictive occupancy models
• Test models in simulation and with real-world data to demonstrate improvements in safety, transparency, and performance
• Publish findings at leading AI and robotics conferences

• Master's degree in Computer Science, Robotics, AI, Data Science, or related field suitable for doctoral studies
• Strong expertise in Deep Learning and Computer Vision
• Hands-on with PyTorch and solid Python programming
• Experience in 3D Computer Vision, Occupancy Networks, End-to-End Driving, World Models, or Sensor Fusion (desirable)
• Familiarity with autonomous driving datasets and simulation environments (e.g., CARLA) (advantage)
• Creative, self-driven, ownership-focused research approach
• Excellent English; German considered an advantage

• creative thinking
• self-driven
• ownership of topics
• Deep Learning
• Computer Vision
• PyTorch