Your tasks
The ideal candidate will passionately drive and further advance innovations in the field of reinforcement learning and simulation for planning in autonomous driving. The work includes conducting advanced research and development of novel algorithms and conducting experiments to bring innovation ideas to products.
As an intern, you will work closely with a team of researchers and engineers on one of the following areas:
- GPU-accelerated Simulation for Reinforcement Learning
Build and optimize high-performance, scalable simulation environments tailored for reinforcement learning in autonomous driving scenarios. - Integration and Development of ML-based Planning Models
Develop, train, and integrate planning models for autonomous driving using GPU-accelerated simulations to validate and enhance performance in complex scenarios. - Imitation + Reinforcement Learning with Self-Play
Develop and improve hybrid learning approaches that combine imitation learning and reinforcement learning, with a focus on multi-agent self-play techniques.
Responsibilities include but not limited to:
- Participate in cutting-edge engineering projects applying deep learning and reinforcement learning to tackle challenges in planning and simulation for autonomous driving contexts.
- Work with an international team of experts to transfer the results of advanced research to Bosch business units. Benchmark, validate and test research ideas on simulated environments, large scale datasets, and self-driving vehicles.
- Collaborate with a team of domain experts on novel approaches to learning-based planning and decision-making.
- Benchmark, validate, and iterate on models using large-scale simulation and datasets.
- Communicate research findings through internal reports and/or external publications.
Basic Qualifications
- Currently pursuing MS or PhD in Computer Science / Robotics / Systems Engineering or a related technical field, with research focus on high-performance simulation, reinforcement learning, robotic systems, or autonomous driving applications.
- Hands-on experience in deep learning and/or AI system topics with focus on at least two of the following areas: reinforcement learning, vector/point-based input representations for learning, planning for navigation, multi-agent training / self-play, and autonomous driving.
- Programming experience in C++, Python, and hands-on experience with libraries such as PyTorch, CUDA, Tensorflow, etc.
- Minimum GPA of 3.0
Preferred Qualifications
- Publication record in top venues in robotics/machine learning/computer vision, e.g., ICRA, IROS, RSS, NeurIPS, ICML, ICLR, CVPR, ICCV, and ECCV.
- Project experience in the field of planning or simulation for automated driving
- Strong leadership skills with excellent English communication & teamwork skills.
- Background in high-performance simulation, reinforcement learning, or machine learning for autonomous driving.
- Background in probabilistic robotics.
- Experience in writing algorithms in C++ efficiently and correctly in a production environment (code reviews, unit tests, etc.)
- Experience with the Madrona engine, GPUDrive, or other GPU accelerated simulation frameworks.
- Knowledge of Linux, and development on Linux systems.
