Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models
Achieving stable and reliable autonomous driving in complex traffic environments while ensuring safety under unpredictable conditions is a critical challenge in autonomous driving technology. To address this issue, this study proposes the Safedrive Dreamer navigation framework, which aims to reduce...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | Alexandria Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1110016824011943 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841524920488558592 |
|---|---|
| author | Haitao Li Tao Peng Bangan Wang Ronghui Zhang Bolin Gao Ningguo Qiao Zhiwei Guan Jiayin Li Tianyu shi |
| author_facet | Haitao Li Tao Peng Bangan Wang Ronghui Zhang Bolin Gao Ningguo Qiao Zhiwei Guan Jiayin Li Tianyu shi |
| author_sort | Haitao Li |
| collection | DOAJ |
| description | Achieving stable and reliable autonomous driving in complex traffic environments while ensuring safety under unpredictable conditions is a critical challenge in autonomous driving technology. To address this issue, this study proposes the Safedrive Dreamer navigation framework, which aims to reduce the reliance on trial-and-error learning in real-world scenarios, thereby mitigating the risks associated with dynamic driving conditions and enhancing vehicle foresight. This framework integrates the predictive capabilities of world models with the constrained Markov decision process (CMDP) and safety reinforcement learning to accurately anticipate future environmental changes. This ensures the reliability of autonomous driving routes, thereby improving both safety and efficiency. Furthermore, to reduce trial-and-error costs in real-world applications, this study employs PAC-Bayesian methods to derive generalization error bounds between simulations and reality, enabling a more effective transfer of knowledge and experience from simulations to real-world scenarios. Validation experiments in simulated and real environments showed that Safedrive Dreamer significantly outperformed existing autonomous driving solutions by 3.8% in key safety metrics, excelling in collision avoidance and risk reduction. This study provides new insights into the integration of world models into decision-making processes to enhance decision-making capabilities in safety–critical applications, thereby contributing significantly to the improvement of autonomous driving safety and reliability. |
| format | Article |
| id | doaj-art-b21c3946ad294341aec41d7559569e2b |
| institution | Kabale University |
| issn | 1110-0168 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Alexandria Engineering Journal |
| spelling | doaj-art-b21c3946ad294341aec41d7559569e2b2025-01-18T05:03:36ZengElsevierAlexandria Engineering Journal1110-01682025-01-0111192106Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world modelsHaitao Li0Tao Peng1Bangan Wang2Ronghui Zhang3Bolin Gao4Ningguo Qiao5Zhiwei Guan6Jiayin Li7Tianyu shi8Tianjin University of Technology and Education, The School of Automotive and Transportation, Tianjin, 300222, Tianjin, China; University of Toronto, Department of Civil and Mineral Engineering, Toronto, M5S 2E8, Ontario, CanadaTianjin University of Technology and Education, The School of Automotive and Transportation, Tianjin, 300222, Tianjin, China; Shanghai Artificial Intelligence Laboratory, Department of Artificial Intelligence, Shanghai, 200232, Shanghai, China; Corresponding author at: Tianjin University of Technology and Education, The School of Automotive and Transportation, Tianjin, 300222, Tianjin, China.University of Toronto, Mechanical Engineering, Toronto, M5S 2E8, Ontario, CanadaSun Yat-sen University, School of Intelligent Systems Engineering, canton, 510275, Guangdong, ChinaTsinghua University, School of Vehicle and Mobility, Beijing, 100084, Beijing, ChinaTianjin University of Technology and Education, The School of Automotive and Transportation, Tianjin, 300222, Tianjin, ChinaTianjin Sino-German University of Applied Sciences, The Automobile and Rail Transportation College, Tianjin, 300350, Tianjin, ChinaXiamen University Malaysia, School of Computing and Data Science, Selangor, 43900, Sepang, MalaysiaUniversity of Toronto, Department of Civil and Mineral Engineering, Toronto, M5S 2E8, Ontario, CanadaAchieving stable and reliable autonomous driving in complex traffic environments while ensuring safety under unpredictable conditions is a critical challenge in autonomous driving technology. To address this issue, this study proposes the Safedrive Dreamer navigation framework, which aims to reduce the reliance on trial-and-error learning in real-world scenarios, thereby mitigating the risks associated with dynamic driving conditions and enhancing vehicle foresight. This framework integrates the predictive capabilities of world models with the constrained Markov decision process (CMDP) and safety reinforcement learning to accurately anticipate future environmental changes. This ensures the reliability of autonomous driving routes, thereby improving both safety and efficiency. Furthermore, to reduce trial-and-error costs in real-world applications, this study employs PAC-Bayesian methods to derive generalization error bounds between simulations and reality, enabling a more effective transfer of knowledge and experience from simulations to real-world scenarios. Validation experiments in simulated and real environments showed that Safedrive Dreamer significantly outperformed existing autonomous driving solutions by 3.8% in key safety metrics, excelling in collision avoidance and risk reduction. This study provides new insights into the integration of world models into decision-making processes to enhance decision-making capabilities in safety–critical applications, thereby contributing significantly to the improvement of autonomous driving safety and reliability.http://www.sciencedirect.com/science/article/pii/S1110016824011943Reinforcement learningWorld modelsReality gapDeep learning |
| spellingShingle | Haitao Li Tao Peng Bangan Wang Ronghui Zhang Bolin Gao Ningguo Qiao Zhiwei Guan Jiayin Li Tianyu shi Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models Alexandria Engineering Journal Reinforcement learning World models Reality gap Deep learning |
| title | Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models |
| title_full | Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models |
| title_fullStr | Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models |
| title_full_unstemmed | Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models |
| title_short | Safedrive dreamer: Navigating safety–critical scenarios in autonomous driving with world models |
| title_sort | safedrive dreamer navigating safety critical scenarios in autonomous driving with world models |
| topic | Reinforcement learning World models Reality gap Deep learning |
| url | http://www.sciencedirect.com/science/article/pii/S1110016824011943 |
| work_keys_str_mv | AT haitaoli safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT taopeng safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT banganwang safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT ronghuizhang safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT bolingao safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT ningguoqiao safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT zhiweiguan safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT jiayinli safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels AT tianyushi safedrivedreamernavigatingsafetycriticalscenariosinautonomousdrivingwithworldmodels |