T4P: Test-Time Training of Trajectory Prediction via Masked Autoencoder and Actor-specific Token Memory
This paper presents a novel trajectory prediction method using masked autoencoders and actor-specific token memory to improve accuracy and efficiency under distribution shifts.
Jun 17, 2024
Improving Transferability for Cross-domain Trajectory Prediction via Neural Stochastic Differential Equation
This paper proposes a method using Neural Stochastic Differential Equations to improve cross-dataset transferability in multi-agent trajectory prediction by addressing discrepancies in data acquisition strategies.
Mar 24, 2024
Leveraging Future Relationship Reasoning for Vehicle Trajectory Prediction
This paper presents a novel vehicle trajectory prediction approach using lane information and probabilistic modeling to predict future agent interactions, achieving state-of-the-art performance on long-term prediction benchmarks.
May 24, 2023