Transfer Learning Study of Motion Transformer-based Trajectory Predictions
This work addresses the challenge of deploying simulation-trained models in real-world autonomous driving, which is incremental as it applies existing transfer learning methods to a specific domain.
The study tackled the problem of adapting transformer-based trajectory prediction models from simulation to real-world autonomous driving, finding that transfer learning techniques can help manage domain shifts but involve trade-offs between computational time and performance.
Trajectory planning in autonomous driving is highly dependent on predicting the emergent behavior of other road users. Learning-based methods are currently showing impressive results in simulation-based challenges, with transformer-based architectures technologically leading the way. Ultimately, however, predictions are needed in the real world. In addition to the shifts from simulation to the real world, many vehicle- and country-specific shifts, i.e. differences in sensor systems, fusion and perception algorithms as well as traffic rules and laws, are on the agenda. Since models that can cover all system setups and design domains at once are not yet foreseeable, model adaptation plays a central role. Therefore, a simulation-based study on transfer learning techniques is conducted on basis of a transformer-based model. Furthermore, the study aims to provide insights into possible trade-offs between computational time and performance to support effective transfers into the real world.