论文
围绕三维感知、协同占用预测与高效智能系统的研究工作。
我的论文与投稿工作围绕一条长期主线展开:面向自动驾驶和具身智能体的高效预测式三维场景理解。目前重点关注协同语义占用预测、带宽感知 token 通信和占用式世界模型;早期工作则来自资源受限嵌入式系统中的实时调度优化。
这些工作共同体现了我希望在 Ph.D. 阶段继续推进的方向:构建准确、通信高效、具备时间建模能力,并且能够服务下游决策的感知系统。
2026
- NeurIPS 2026Learning to Merge Tokens for Communication-Efficient Collaborative Occupancy PredictionSubmitted, 2026
- WACV 2027CAMerge: Context-Aware Token Merging for Efficient Semantic SegmentationSubmitted, 2026
- AAAI 2027FedHarness: Federated Harness Evolution for LLM AgentsSubmitted, 2026
- AAAI 2027CriticShift: Reliability-Aware Tool-Output Selection for LLM AgentsSubmitted, 2026
- ITSC 2026Bandwidth-Aware Adaptive Token Communication for Collaborative Occupancy PredictionSubmitted, 2026
- In Prep.Collaborative 4D Occupancy World Models with Motion-Aware Token MemoryManuscript in preparation, 2026
2023
- AIKIIEScheduling Optimization Design of IoT Embedded System Based on Improved RMS AlgorithmIn International Conference on Ambient Intelligence, Knowledge Informatics and Industrial Electronics, 2023