PhD Domain Adaptation and Basic Models for Automated Driving Perception

Company Description

Do you want beneficial technologies to be shaped by your ideas? Whether in the areas of mobility solutions, consumer goods, industrial technology, or energy and building technology – with us, you will have the chance to improve the quality of life across the globe. Welcome to Bosch.

The Robert Bosch GmbH is looking forward to your application!

Job Description

Unsupervised Domain Adaptation techniques have become increasingly important in the context of automated driving, as they address the challenge of adapting perception models to new and diverse driving environments. They aim to learn a shared representation space that minimizes the discrepancy between the source and target domains. By doing so, these techniques enable the deployment of automated driving systems in various real-world scenarios including different weather conditions and road types. Developing and improving these techniques is critical to realizing the full potential of autonomous driving and advancing the field of automated driving perception, especially when labeled data in the target domain is not available or difficult to obtain.

In addition, foundational models have the potential to serve as the bedrock upon which the adaptability and effectiveness of machine learning systems are built. These models are designed to learn a robust representation of data that transcends domain-specific peculiarities, allowing for seamless transfer of knowledge and insights across diverse domains. By capturing the fundamental underlying patterns and features common to different datasets, foundational models enable the adaptation of machine learning algorithms to new environments, making them an essential component in addressing challenges such as domain shift and data scarcity. In essence, foundational models pave the way for more versatile and adaptive artificial intelligence systems capable of performing reliably across a wide spectrum of real-world scenarios.

Your Tasks:

• Investigate and develop state-of-the-art Unsupervised Domain Adaptation and Foundational Model frameworks with a focus on point-cloud based sensors, e.g., Radar & Lidar.
• Evaluation of the developed methods on both open-source and internal proprietary datasets, both offline and online.
• Work alongside experienced researchers and developers to contribute to and support large-scale open-source projects in the field of automated driving.
• Publication of research findings in top-tier conferences and journals (CVPR, ICML, NIPS, ICLR or IV)
• Supervision of Master's and Bachelor's students
• Take ownership and responsibility for project tasks and work in an agile and diverse team with exchange across several projects and groups

Qualifications

• Education: degree (Master/Diploma) in Computer Science, Electrical Engineering, Mathematics or related field with excellent academic achievements
• Experience and Knowledge: solid knowledge of machine learning algorithms and principles, preferably deep learning. Prior experience in domain adaptation is a plus.
• Programming Experience: Good proficiency in Python with familiarity with deep learning frameworks such as Tensorflow or PyTorch
• Personality and Working Practice: Open-minded team player who is goal-oriented with excellent problem-solving and analytical skills
• Language: fluent in English (written and spoken), German is a plus

Additional Information

The final Ph.D. topic is subject to your university. Start: according to prior agreement

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

                                    Need help with your application?
Claudia Schillerwein (Human Resources)
[email protected]

Need more information about the job?
Karim Armanious (Functional Department)                                              [email protected]