arXiv subjects

Tung Phan-Minh

Publications and source records attributed to Tung Phan-Minh.

9 recordsLinked to original sources

arXiv2026

Explainable deep learning improves human mental models of self-driving cars

Self-driving cars increasingly rely on deep neural networks to achieve human-like driving. The opacity of such black-box planners makes it challenging to accurately anticipate when they will fail, with potentially catastrophic consequences. While research into interpreting these systems has surged, most of it is confined to simulations or toy setups due to the difficulty of real-world deployment, leaving the practical utility of such techniques unknown. Here, we introduce the Concept-Wrapper Network (CW-Net), a method for faithfully explaining the behavior of machine-learning-based planners that causally grounds their reasoning in human-interpretable concepts without sacrificing performance. We deploy CW-Net on a real self-driving car and show that the resulting explanations improve the human driver's mental model of the vehicle, allowing them to better predict its behavior, particularly in surprising situations. This demonstrates that explainable deep learning integrated into self-driving cars can be both understandable and useful in a realistic deployment setting. We anticipate our method could be applied to other safety-critical systems, such as autonomous drones and robotic surgeons, as well as to other architectures, such as end-to-end learning systems and vision-language-action models. Overall, our study establishes a deployment-validated pathway to interpretability for autonomous agents, which could help make them more transparent and safe.

cs.RO↗

arXiv2025

Causal Composition Diffusion Model for Closed-loop Traffic Generation

Simulation is critical for safety evaluation in autonomous driving, particularly in capturing complex interactive behaviors. However, generating realistic and controllable traffic scenarios in long-tail situations remains a significant challenge. Existing generative models suffer from the conflicting objective between user-defined controllability and realism constraints, which is amplified in safety-critical contexts. In this work, we introduce the Causal Compositional Diffusion Model (CCDiff), a structure-guided diffusion framework to address these challenges. We first formulate the learning of controllable and realistic closed-loop simulation as a constrained optimization problem. Then, CCDiff maximizes controllability while adhering to realism by automatically identifying and injecting causal structures directly into the diffusion process, providing structured guidance to enhance both realism and controllability. Through rigorous evaluations on benchmark datasets and in a closed-loop simulator, CCDiff demonstrates substantial gains over state-of-the-art approaches in generating realistic and user-preferred trajectories. Our results show CCDiff's effectiveness in extracting and leveraging causal structures, showing improved closed-loop performance based on key metrics such as collision rate, off-road rate, FDE, and comfort.

cs.AI↗

arXiv2025

DriveGPT: Scaling Autoregressive Behavior Models for Driving

We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.

cs.LG↗

arXiv2022

Driving in Real Life with Inverse Reinforcement Learning

In this paper, we introduce the first learning-based planner to drive a car in dense, urban traffic using Inverse Reinforcement Learning (IRL). Our planner, DriveIRL, generates a diverse set of trajectory proposals, filters these trajectories with a lightweight and interpretable safety filter, and then uses a learned model to score each remaining trajectory. The best trajectory is then tracked by the low-level controller of our self-driving vehicle. We train our trajectory scoring model on a 500+ hour real-world dataset of expert driving demonstrations in Las Vegas within the maximum entropy IRL framework. DriveIRL's benefits include: a simple design due to only learning the trajectory scoring function, relatively interpretable features, and strong real-world performance. We validated DriveIRL on the Las Vegas Strip and demonstrated fully autonomous driving in heavy traffic, including scenarios involving cut-ins, abrupt braking by the lead vehicle, and hotel pickup/dropoff zones. Our dataset will be made public to help further research in this area.

cs.RO↗

arXiv2021

Failure-Tolerant Contract-Based Design of an Automated Valet Parking System using a Directive-Response Architecture

Increased complexity in cyber-physical systems calls for modular system design methodologies that guarantee correct and reliable behavior, both in normal operations and in the presence of failures. This paper aims to extend the contract-based design approach using a directive-response architecture to enable reactivity to failure scenarios. The architecture is demonstrated on a modular automated valet parking (AVP) system. The contracts for the different components in the AVP system are explicitly defined, implemented, and validated against a Python implementation.

eess.SY↗

arXiv2021

Rules of the Road: Safety and Liveness Guarantees for Autonomous Vehicles

The ability to guarantee safety and progress for all vehicles is vital to the success of the autonomous vehicle industry. We present a framework for designing autonomous vehicle behavior in a way that is safe and guarantees progress for all agents. In this paper, we first introduce a new game paradigm which we term the quasi-simultaneous game. We then define an agent protocol that all agents must use to make decisions in this quasi-simultaneous game setting. According to the protocol, agents first select an intended action using a behavioral profile. Then, the protocol defines whether an agent has precedence to take its intended action or must take a sub-optimal action. The protocol ensures safety under all traffic conditions and liveness for all agents under `sparse' traffic conditions. We provide proofs of correctness of the protocol and validate our results in simulation.

cs.MA↗

arXiv2020

CoverNet: Multimodal Behavior Prediction using Trajectory Sets

We present CoverNet, a new method for multimodal, probabilistic trajectory prediction for urban driving. Previous work has employed a variety of methods, including multimodal regression, occupancy maps, and 1-step stochastic policies. We instead frame the trajectory prediction problem as classification over a diverse set of trajectories. The size of this set remains manageable due to the limited number of distinct actions that can be taken over a reasonable prediction horizon. We structure the trajectory set to a) ensure a desired level of coverage of the state space, and b) eliminate physically impossible trajectories. By dynamically generating trajectory sets based on the agent's current state, we can further improve our method's efficiency. We demonstrate our approach on public, real-world self-driving datasets, and show that it outperforms state-of-the-art methods.

cs.LG↗

arXiv2020

Counter-example Guided Learning of Bounds on Environment Behavior

There is a growing interest in building autonomous systems that interact with complex environments. The difficulty associated with obtaining an accurate model for such environments poses a challenge to the task of assessing and guaranteeing the system's performance. We present a data-driven solution that allows for a system to be evaluated for specification conformance without an accurate model of the environment. Our approach involves learning a conservative reactive bound of the environment's behavior using data and specification of the system's desired behavior. First, the approach begins by learning a conservative reactive bound on the environment's actions that captures its possible behaviors with high probability. This bound is then used to assist verification, and if the verification fails under this bound, the algorithm returns counter-examples to show how failure occurs and then uses these to refine the bound. We demonstrate the applicability of the approach through two case-studies: i) verifying controllers for a toy multi-robot system, and ii) verifying an instance of human-robot interaction during a lane-change maneuver given real-world human driving data.

eess.SY↗

arXiv2019

Towards Assume-Guarantee Profiles for Autonomous Vehicles

Rules or specifications for autonomous vehicles are currently formulated on a case-by-case basis, and put together in a rather ad-hoc fashion. As a step towards eliminating this practice, we propose a systematic procedure for generating a set of supervisory specifications for self-driving cars that are 1) associated with a distributed assume-guarantee structure and 2) characterizable by the notion of consistency and completeness. Besides helping autonomous vehicles make better decisions on the road, the assume-guarantee contract structure also helps address the notion of blame when undesirable events occur. We give several game-theoretic examples to demonstrate applicability of our framework.

eess.SY↗