Building a deep learning framework for driving behavior prediction with data fusion.

Context

With the development of autonomous driving technology, driving behavior prediction has become an important part of this field, which is crucial for safe and efficient driving. Here, relevant data is collected from various sources, such as the cloud and real-world driving scenarios. This data is then integrated into models to train different neural networks, enabling more accurate and efficient driving behavior predictions.

Tasks

• Develop a GUI to visualize the collected data from the cloud.
• Literature research on the SOTA technique
• Comparison of SOTA models for driving behavior prediction, analysis of the different influencing factors, such as driving style, driving experience, traffic safety, etc.
• Development, training and optimization of neural networks for short and long term driving behavior prediction.

Requirements

• Motivation and interest in solving technical problems independently
• Experience with programming (Python, C++ , Java...)
• Knowledge of machine learning, ideally reinforcement learning and data fusion
• Experience with deep learning frameworks
• Analytical, problem solving and communication skills