MVAdapt: Zero-Shot Multi-Vehicle Adaptation for End-to-End Autonomous Driving

2026-04-13

AI Summary

This work addresses the performance degradation of end-to-end autonomous driving models when deployed across vehicles with differing dynamics. To tackle this challenge, the authors propose MVAdapt, a novel framework that explicitly models vehicle physical characteristics as conditional information. By integrating a lightweight physics encoder with a cross-attention mechanism, MVAdapt injects this physical prior into frozen scene representations, enabling zero-shot policy transfer across vehicle types without retraining the backbone network. Built upon the TransFuser++ architecture and evaluated in the CARLA simulation environment, MVAdapt significantly outperforms baseline methods on the CARLA Leaderboard 1.0. It not only supports zero-shot adaptation to multiple unseen vehicle configurations but also efficiently calibrates policies for vehicles with extreme physical discrepancies using minimal additional data.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

End-to-End (E2E) autonomous driving models are usually trained and evaluated with a fixed ego-vehicle, even though their driving policy is implicitly tied to vehicle dynamics. When such a model is deployed on a vehicle with different size, mass, or drivetrain characteristics, its performance can degrade substantially; we refer to this problem as the vehicle-domain gap. To address it, we propose MVAdapt, a physics-conditioned adaptation framework for multi-vehicle E2E driving. MVAdapt combines a frozen TransFuser++ scene encoder with a lightweight physics encoder and a cross-attention module that conditions scene features on vehicle properties before waypoint decoding. In the CARLA Leaderboard 1.0 benchmark, MVAdapt improves over naive transfer and multi-embodiment adaptation baselines on both in-distribution and unseen vehicles. We further show two complementary behaviors: strong zero-shot transfer on many unseen vehicles, and data-efficient few-shot calibration for severe physical outliers. These results suggest that explicitly conditioning E2E driving policies on vehicle physics is an effective step toward more transferable autonomous driving models. All codes are available at https://github.com/hae-sung-oh/MVAdapt

Problem

• End-to-End autonomous driving
• vehicle-domain gap
• multi-vehicle adaptation
• zero-shot transfer
• vehicle dynamics

Innovation

• zero-shot adaptation
• physics-conditioned driving
• multi-vehicle transfer
• end-to-end autonomous driving
• cross-attention conditioning