5 accepted papers
We present DIO, a flexible world model that can estimate the scene occupancy-flow from a sparse set of LiDAR observations, and decompose it into individual instances. DIO can not only complete instance shapes at the present time, but also forecast their occupancy-flow evolution over a future horizon…
Distribution shifts between operational domains can severely affect the performance of learned models in self-driving vehicles (SDVs). While this is a well-established problem, prior work has mostly explored naive solutions such as fine-tuning, focusing on the motion prediction task. In this work, w
This work uses game theory as a mathematical framework to address interaction modeling in multi-agent motion forecasting and control. Despite its interpretability, applying game theory to real-world robotics, like automated driving, faces challenges such as unknown game parameters. To tackle these,…
Offline reinforcement learning (RL) provides a framework for learning decision-making from offline data and therefore constitutes a promising approach for real-world applications such as automated driving (AD). Especially in safety-critical applications, interpretability and transferability are cruc
It is widely assumed that considering vehicle interactions for trajectory prediction can significantly improve accuracy. All environmental sensors of an automated vehicle (AV) suffer from occlusion. Therefore, relevant vehicles can be occluded by others, especially in dense traffic situations, and r