arXiv subjects

Hongbo Gao

Publications and source records attributed to Hongbo Gao.

9 recordsLinked to original sources

arXiv2026

AGM: Achievement-Grounded Memory for Closed-Loop Agents with Frozen VLA Policies

Frozen vision-language-action (VLA) policies offer broad manipulation skills but execute open-loop action chunks without tracking task progress, so the agent cannot reliably decide whether to continue, retry, or terminate. External memory is a natural remedy, yet it can be harmful when attempted actions are treated as completed progress, turning local execution errors into persistent task-state errors. We propose Achievement-Grounded Memory (AGM), a lightweight closed-loop framework for frozen VLA policies that represents a task as a subgoal sequence with a progress pointer and advances this memory only after the current subgoal is verified by physical evidence. Proprioceptive interaction cues decide when to verify, while coherent point tracking and language-conditioned cross-view comparison, sourced from frozen foundation models through a single 2.43M-parameter verification head, decide what was achieved. AGM thereby converts open-loop execution into a closed loop of execution, verification, and progress, keeping the policy frozen without test-time large-model inference. On the RoboMME Counting benchmark, AGM reaches on PickXTimes and on BinFill, surpassing the strongest memory-augmented baseline by points on average, and the framework yields equally decisive gains on a physical robot. Reliable embodied memory thus depends more on disciplined state updates than on memory capacity.

cs.RO↗

arXiv2026

OccamView: Object-Conditioned View Selection for Frame-Budgeted Active 3D Gaussian Reconstruction

Active 3D Gaussian reconstruction fundamentally relies on selecting informative next-best views under limited sensing budgets. Existing active 3DGS methods primarily plan viewpoints according to geometric information gain, treating object-induced hidden regions in the same manner as general unexplored space. Under tight frame budgets, such geometry-driven strategies may prioritize global scene coverage while leaving partially observed objects incompletely reconstructed. To address this limitation, we propose OccamView, an object-conditioned view-selection framework for frame-budgeted active 3D Gaussian reconstruction. Rather than predicting unseen object geometry or performing shape completion, OccamView maintains an online object memory from open-vocabulary detections grounded in measured RGB-D observations and represents unresolved local occupancy around detected objects as conservative hidden-region proxies. Candidate viewpoints are then evaluated using an occlusion-aware proxy-coverage score. Furthermore, we introduce a Geo-Floor mechanism that restricts object-conditioned re-ranking to geometrically competitive candidates, allowing object-conditioned cues to guide complementary observations while preserving the geometry-driven exploration behavior of the underlying planner. Experiments on Replica and Matterport3D under a unified frame-budgeted protocol show that OccamView consistently reduces Completion and improves Completion Ratio across five frame budgets, with particularly pronounced gains under limited frame budgets. These results demonstrate that lightweight object-conditioned cues effectively complement geometry-driven active view planning.

cs.RO↗

arXiv2026

Eliminating Propagation Delay: Attention-Based Spatial-Temporal Fusion Graph Convolution Network for Traffic Flow Prediction

Predicting traffic flow is crucial to optimizing transportation systems and improving urban mobility. Many graph convolution-based models have been proposed to extract spatial-temporal features and predict traffic flow. However, most focus on spatial-temporal and semantic correlation in topological relationships. There are two primary problems to address. Firstly, the convolutional structure in the model focuses on utilizing static spatial dependencies and spatial-temporal relationships in topological structures, while neglecting the different information propagation delays between adjacent nodes in the convolution. Secondly, these methods often stack a large number of complex structures, resulting in a substantial increase in computational time during the model training phase, thereby disregarding the model's requirements for timeliness. In this paper, we propose a novel network called the Attention-Based Spatial-Temporal Fusion Graph Convolution Network (A-STFGCN). We design a spatial-temporal fusion block to extract the spatial-temporal feature correlations with propagation delay errors removed and to capture both long-term and short-term temporal characteristics of the data within a multi-head self-attention mechanism based on a mask matrix. Extensive experiments on five real-world datasets demonstrate that our method achieves the best overall performance while having good computation and data utilization efficiency compared with the eight baseline methods.

cs.LG↗

arXiv2025

STEP: Stepwise Curriculum Learning for Context-Knowledge Fusion in Conversational Recommendation

Conversational recommender systems (CRSs) aim to proactively capture user preferences through natural language dialogue and recommend high-quality items. To achieve this, CRS gathers user preferences via a dialog module and builds user profiles through a recommendation module to generate appropriate recommendations. However, existing CRS faces challenges in capturing the deep semantics of user preferences and dialogue context. In particular, the efficient integration of external knowledge graph (KG) information into dialogue generation and recommendation remains a pressing issue. Traditional approaches typically combine KG information directly with dialogue content, which often struggles with complex semantic relationships, resulting in recommendations that may not align with user expectations. To address these challenges, we introduce STEP, a conversational recommender centered on pre-trained language models that combines curriculum-guided context-knowledge fusion with lightweight task-specific prompt tuning. At its heart, an F-Former progressively aligns the dialogue context with knowledge-graph entities through a three-stage curriculum, thus resolving fine-grained semantic mismatches. The fused representation is then injected into the frozen language model via two minimal yet adaptive prefix prompts: a conversation prefix that steers response generation toward user intent and a recommendation prefix that biases item ranking toward knowledge-consistent candidates. This dual-prompt scheme allows the model to share cross-task semantics while respecting the distinct objectives of dialogue and recommendation. Experimental results show that STEP outperforms mainstream methods in the precision of recommendation and dialogue quality in two public datasets.

cs.AI↗

arXiv2024

Pretraining-finetuning Framework for Efficient Co-design: A Case Study on Quadruped Robot Parkour

In nature, animals with exceptional locomotion abilities, such as cougars, often possess asymmetric fore and hind legs. This observation inspired us: could optimizing the leg length of quadruped robots endow them with similar locomotive capabilities? In this paper, we propose an approach that co-optimizes the mechanical structure and control policy to boost the locomotive prowess of quadruped robots. Specifically, we introduce a novel pretraining-finetuning framework, which not only guarantees optimal control strategies for each mechanical candidate but also ensures time efficiency. Additionally, we have devised an innovative training method for our pretraining network, integrating spatial domain randomization with regularization methods, markedly improving the network's generalizability. Our experimental results indicate that the proposed pretraining-finetuning framework significantly enhances the overall co-design performance with less time consumption. Moreover, the co-design strategy substantially exceeds the conventional method of independently optimizing control strategies, further improving the robot's locomotive performance and providing an innovative approach to enhancing the extreme parkour capabilities of quadruped robots.

cs.RO↗

arXiv2024

Pre-trained Transformer-Enabled Strategies with Human-Guided Fine-Tuning for End-to-end Navigation of Autonomous Vehicles

Autonomous driving (AD) technology, leveraging artificial intelligence, strives for vehicle automation. End-toend strategies, emerging to simplify traditional driving systems by integrating perception, decision-making, and control, offer new avenues for advanced driving functionalities. Despite their potential, current challenges include data efficiency, training complexities, and poor generalization. This study addresses these issues with a novel end-to-end AD training model, enhancing system adaptability and intelligence. The model incorporates a Transformer module into the policy network, undergoing initial behavior cloning (BC) pre-training for update gradients. Subsequently, fine-tuning through reinforcement learning with human guidance (RLHG) adapts the model to specific driving environments, aiming to surpass the performance limits of imitation learning (IL). The fine-tuning process involves human interactions, guiding the model to acquire more efficient and safer driving behaviors through supervision, intervention, demonstration, and reward feedback. Simulation results demonstrate that this framework accelerates learning, achieving precise control and significantly enhancing safety and reliability. Compared to other advanced baseline methods, the proposed approach excels in challenging AD tasks. The introduction of the Transformer module and human-guided fine-tuning provides valuable insights and methods for research and applications in the AD field.

cs.RO↗

arXiv2023

Integrated Drill Boom Hole-Seeking Control via Reinforcement Learning

Intelligent drill boom hole-seeking is a promising technology for enhancing drilling efficiency, mitigating potential safety hazards, and relieving human operators. Most existing intelligent drill boom control methods rely on a hierarchical control framework based on inverse kinematics. However, these methods are generally time-consuming due to the computational complexity of inverse kinematics and the inefficiency of the sequential execution of multiple joints. To tackle these challenges, this study proposes an integrated drill boom control method based on Reinforcement Learning (RL). We develop an integrated drill boom control framework that utilizes a parameterized policy to directly generate control inputs for all joints at each time step, taking advantage of joint posture and target hole information. By formulating the hole-seeking task as a Markov decision process, contemporary mainstream RL algorithms can be directly employed to learn a hole-seeking policy, thus eliminating the need for inverse kinematics solutions and promoting cooperative multi-joint control. To enhance the drilling accuracy throughout the entire drilling process, we devise a state representation that combines Denavit-Hartenberg joint information and preview hole-seeking discrepancy data. Simulation results show that the proposed method significantly outperforms traditional methods in terms of hole-seeking accuracy and time efficiency.

cs.RO↗

arXiv2021

Receding Horizon Control Based Online Motion Planning with Partially Infeasible LTL Specifications

This work considers online optimal motion planning of an autonomous agent subject to linear temporal logic (LTL) constraints. The environment is dynamic in the sense of containing mobile obstacles and time-varying areas of interest (i.e., time-varying reward and workspace properties) to be visited by the agent. Since user-specified tasks may not be fully realized (i.e., partially infeasible), this work considers hard and soft LTL constraints, where hard constraints enforce safety requirement (e.g. avoid obstacles) while soft constraints represent tasks that can be relaxed to not strictly follow user specifications. The motion planning of the agent is to generate policies, in decreasing order of priority, to 1) formally guarantee the satisfaction of safety constraints; 2) mostly satisfy soft constraints (i.e., minimize the violation cost if desired tasks are partially infeasible); and 3) optimize the objective of rewards collection (i.e., visiting dynamic areas of more interests). To achieve these objectives, a relaxed product automaton, which allows the agent to not strictly follow the desired LTL constraints, is constructed. A utility function is developed to quantify the differences between the revised and the desired motion plan, and the accumulated rewards are designed to bias the motion plan towards those areas of more interests. Receding horizon control is synthesized with an LTL formula to maximize the accumulated utilities over a finite horizon, while ensuring that safety constraints are fully satisfied and soft constraints are mostly satisfied. Simulation and experiment results are provided to demonstrate the effectiveness of the developed motion strategy.

cs.RO↗

arXiv2019

The Effects of Using Taxi-Hailing Application on Driving Performance

Driver distraction has become a major threat to the road safety, and the globally booming taxi-hailing application introduces new source of distraction to drivers. Although various in-vehicle information systems (IVIS) have been studied extensively, no documentation exists objectively measuring the extent to which interacting with taxi-hailing application during driving impacts drivers' behavior. To fill this gap, a simulator-based study was conducted to synthetically compare the effects that different output modalities (visual, audio, combined visual-audio) and input modalities (baseline, manual, speech) imposed on the driving performance. The results show that the visual output introduced more negative effects on driving performance compared to audio output. In the combined output, visual component dominated the effects imposed on the longitudinal control and hazard detection; audio component only exacerbated the negative effects of visual component on the lateral control. Speech input modality was overall less detrimental to driving performance than manual input modality, especially reflected in the drivers' quicker reaction to hazard events. The visual-manual interaction modality most severely impaired the hazard detecting ability, while also led to strong compensative behaviors. The audio-speech and visual-speech modality associated with more smooth lateral control and faster response to hazard events respectively compared to other modality. These results could be applied to improve the design of not only the taxi-hailing application, but also other input-output balanced IVIS.

cs.HC↗