mlsanityData to 14 Oct 2026

Controllable Generative Trajectory Prediction via Weak Preference Alignment

arXiv 2510.10731 ↗First posted 12 Oct 2025cs.RO cs.LGcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Reward modeling and RLHF

Abstract (arXiv)

Deep generative models such as conditional variational autoencoders (CVAEs) have shown great promise for predicting trajectories of surrounding agents in autonomous vehicle planning. State-of-the-art models have achieved remarkable accuracy in such prediction tasks. Besides accuracy, diversity is also crucial for safe planning because human behaviors are inherently uncertain and multimodal. However, existing methods generally lack a scheme to generate controllably diverse trajectories, which is arguably more useful than randomly diversified trajectories, to the end of safe planning. To address this, we propose PrefCVAE, an augmented CVAE framework that uses weakly labeled preference pairs to imbue latent variables with semantic attributes. Using average velocity as an example attribute, we demonstrate that PrefCVAE enables controllable, semantically meaningful predictions without degrading baseline accuracy. Our results show the effectiveness of preference supervision as a cost-effective way to enhance sampling-based generative models.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 26 of its 35 references that are in the map. First posted on arXiv 12 Oct 2025.

Builds on

26 of 35 references.

Closest ancestors · 3

Same direction, newest first

  1. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Model29 May
  2. 2018Reward learning from human preferences and demonstrations in Atari15 Nov
  3. 2017Deep reinforcement learning from human preferences12 Jun

Other references · 23

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,523
  3. 2015Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)cited by 1,070
  4. 2018Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representationscited by 941
  5. 2014Semi-Supervised Learning with Deep Generative Modelscited by 496
  6. 2021AgentFormer: Agent-Aware Transformers for Socio-Temporal Multi-Agent Forecastingcited by 331
  7. 2020Weakly-Supervised Disentanglement Without Compromisescited by 239
  8. 2023MTR++: Multi-Agent Motion Prediction with Symmetric Scene Modeling and Guided Intention Queryingcited by 140
  9. 2017Learning Disentangled Representations with Semi-Supervised Deep Generative Modelscited by 72
  10. 2020Weakly Supervised Disentangled Generative Causal Representation Learningcited by 68
  11. 2019DiversityGAN: Diversity-Aware Vehicle Motion Prediction via Latent Semantic Samplingcited by 40
  12. 2017Latent Constraints: Learning to Generate Conditionally from Unconditional Generative Modelscited by 39
11 more
  1. 2021Variational Autoencoder-Based Vehicle Trajectory Prediction with an Interpretable Latent Spacecited by 21
  2. 2021Injecting Planning-Awareness into Prediction and Detection Evaluationcited by 21
  3. 2022Controllable Data Generation by Deep Learning: A Reviewcited by 20
  4. 2024Learning Semantic Latent Directions for Accurate and Controllable Human Motion Predictioncited by 14
  5. 2019GraphRQI: Classifying Driver Behaviors Using Graph Spectrumscited by 9
  6. 2021TIP: Task-Informed Motion Prediction for Intelligent Vehiclescited by 8
  7. 2024Improving Semantic Control in Discrete Latent Spaces with Transformer Quantized Variational Autoencoderscited by 6
  8. 2023Smooth-Trajectron++: Augmenting the Trajectron++ behaviour prediction model with smooth attentioncited by 1
  9. 2023Controlling the Output of a Generative Model by Latent Feature Vector Shiftingcited by 1
  10. 2024Structural Disentanglement of Causal and Correlated Conceptscited by 1
  11. 2024Critical Example Mining for Vehicle Trajectory Prediction using Flow-based Generative Modelscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Improving Consistency in Vehicle Trajectory Prediction Through Preference Optimization
  2. 2025Generative Models in Decision Making: A Survey
  3. 2025Learning Personalized Driving Styles via Reinforcement Learning from Human Feedback
  4. 2026WALT: Learning World-Model-Aligned Latent Trajectories for Autonomous Driving
  5. 2025Efficient Virtuoso: A Latent Diffusion Transformer Model for Goal-Conditioned Trajectory Planning
  6. 2026ManeuverFlow: Maneuver Hypothesis Generation with Future Evidence for Trajectory Evaluation and Refinement in End-to-End Autonomous Driving
  7. 2026PrefPI: Preference-Guided Steering into Out-of-Distribution Behaviors
  8. 2026MeanFuser: Fast One-Step Multi-Modal Trajectory Generation and Adaptive Reconstruction via MeanFlow for End-to-End Autonomous Driving