Aifen Sui

Hongkuan Zhou, Wei Cao, Aifen Sui et al.

End-to-end autonomous driving, where the entire driving pipeline is replaced with a single neural network, has recently gained research attention because of its simpler structure and faster inference time. Despite this appealing approach largely reducing the complexity in the driving pipeline, it also leads to safety issues because the trained policy is not always compliant with the traffic rules. In this paper, we proposed P-CSG, a penalty-based imitation learning approach with contrastive-based cross semantics generation sensor fusion technologies to increase the overall performance of end-to-end autonomous driving. In this method, we introduce three penalties - red light, stop sign, and curvature speed penalty to make the agent more sensitive to traffic rules. The proposed cross semantics generation helps to align the shared information of different input modalities. We assessed our model's performance using the CARLA Leaderboard - Town 05 Long Benchmark and Longest6 Benchmark, achieving 8.5% and 2.0% driving score improvement compared to the baselines. Furthermore, we conducted robustness evaluations against adversarial attacks like FGSM and Dot attacks, revealing a substantial increase in robustness compared to other baseline models. More detailed information can be found at https://hk-zh.github.io/p-csg-plus.

Hongkuan Zhou, Aifen Sui, Letian Shi et al.

In recent times, there has been a growing focus on end-to-end autonomous driving technologies. This technology involves the replacement of the entire driving pipeline with a single neural network, which has a simpler structure and faster inference time. However, while this approach reduces the number of components in the driving pipeline, it also presents challenges related to interpretability and safety. For instance, the trained policy may not always comply with traffic rules, and it is difficult to determine the reason for such misbehavior due to the lack of intermediate outputs. Additionally, the successful implementation of autonomous driving technology heavily depends on the reliable and expedient processing of sensory data to accurately perceive the surrounding environment. In this paper, we provide penalty-based imitation learning approach combined with cross semantics generation sensor fusion technologies (P-CSG) to efficiently integrate multiple modalities of information and enable the autonomous agent to effectively adhere to traffic regulations. Our model undergoes evaluation within the Town 05 Long benchmark, where we observe a remarkable increase in the driving score by more than 12% when compared to the state-of-the-art (SOTA) model, InterFuser. Notably, our model achieves this performance enhancement while achieving a 7-fold increase in inference speed and reducing the model size by approximately 30%. For more detailed information, including code-based resources, they can be found at https://hk-zh.github.io/p-csg/

Ioannis Tzachristas, Aifen Sui

We introduce a neuro-symbolic framework for multi-intent understanding in mobile AI agents by integrating a structured intent ontology with compact language models. Our method leverages retrieval-augmented prompting, logit biasing and optional classification heads to inject symbolic intent structure into both input and output representations. We formalize a new evaluation metric-Semantic Intent Similarity (SIS)-based on hierarchical ontology depth, capturing semantic proximity even when predicted intents differ lexically. Experiments on a subset of ambiguous/demanding dialogues of MultiWOZ 2.3 (with oracle labels from GPT-o3) demonstrate that a 3B Llama model with ontology augmentation approaches GPT-4 accuracy (85% vs 90%) at a tiny fraction of the energy and memory footprint. Qualitative comparisons show that ontology-augmented models produce more grounded, disambiguated multi-intent interpretations. Our results validate symbolic alignment as an effective strategy for enabling accurate and efficient on-device NLU.