Sampled-Data Hamilton-Jacobi Reachability Based Safe Motion
发布时间:2021-11-28 11:33
随着机器人系统的广泛应用,如何在不确定条件下获取安全的运动轨迹是其面临的重要问题。传统鲁棒运动规划(Motion Planning,MP)不能保证机器人系统运行的性能和鲁棒安全性。对于具有不确定性,状态约束和安全性要求的系统,可达性分析是解决其控制问题的有效方法,但未被广泛应用到MP问题求解中。因此,本文基于可达性分析提出了一套系统的方法,用以解决不确定环境下的鲁棒MP问题。首先,本文进行了静态环境下MP问题的求解。离线部分,通过采样数据(Sampled Data,SD)汉密尔顿-雅可比(Hamilton-Jacobi,HJ)可达性分析方法计算得到系统运动轨迹,然后基于递归算法将原始问题分解成为一系列微分博弈问题。在线部分,给出了一种具有鲁棒安全性和目标可达的反馈控制方法。第二,通过用时间维度对状态进行扩充,本文将上述方法在时变系统中进行了测试。第三,为了解决对抗性环境下的鲁棒MP问题,本文将上述算法引入到高维的联合状态空间下进行推理,并证明了在不失有效性的情况下,方法的维数灾可以通过向低维空间映射得到解决。第四,通过将算法转换到组态空间,我们讨论了其在非质点机器人系统中的应用。本文提...
【文章来源】:上海交通大学上海市 211工程院校 985工程院校 教育部直属院校
【文章页数】:98 页
【学位级别】:硕士
【文章目录】:
摘要
Abstract
Chapter 1 Introduction
1.1 Motivation
1.2 Reachability Analysis
1.3 Related work and contributions
1.4 Methodology and content
Chapter 2 Motion Planning in Static Environments
2.1 Motion Planning as a Reach-Avoid problem
2.2 Sampled-Data Hamilton-Jacobi Reachability (Oine)
2.3 Control synthesis (Online)
2.4 Simulations
Chapter 3 Motion Planning in Time-Varying Environments
3.1 Extension of the Reach-Avoid problem to the State-Time space
3.2 SD HJ Reachability in the State-Time Space (Oine)
3.3 Extended Control Synthesis (Online)
3.4 Simulations
Chapter 4 Motion Planning in Adversarial Environments
4.1 Reach-Avoid Problem in Joint State-Space
4.2 SD HJ Reachability in the Joint State Space (Oine and Online)
4.3 Mitigating the curse of dimensionality using Hamilton-Jacobi projections
4.4 Simulations
Chapter 5 Extension to C-space and case study with the 2-Do F robot arm
5.1 Problem formulation
5.2 Reach-Avoid Problem and SD HJ Reachability in the CT -space
5.3 Projective under-approximation of the SD reach-avoid tube
5.4 Case study with the 2-Do F robot arm
Chapter 6 Conclusion and Future Prospects
6.1 Summary of results
6.2 Future prospects
Acknowledgements
Publications
Bibliography
本文编号:3524349
【文章来源】:上海交通大学上海市 211工程院校 985工程院校 教育部直属院校
【文章页数】:98 页
【学位级别】:硕士
【文章目录】:
摘要
Abstract
Chapter 1 Introduction
1.1 Motivation
1.2 Reachability Analysis
1.3 Related work and contributions
1.4 Methodology and content
Chapter 2 Motion Planning in Static Environments
2.1 Motion Planning as a Reach-Avoid problem
2.2 Sampled-Data Hamilton-Jacobi Reachability (Oine)
2.3 Control synthesis (Online)
2.4 Simulations
Chapter 3 Motion Planning in Time-Varying Environments
3.1 Extension of the Reach-Avoid problem to the State-Time space
3.2 SD HJ Reachability in the State-Time Space (Oine)
3.3 Extended Control Synthesis (Online)
3.4 Simulations
Chapter 4 Motion Planning in Adversarial Environments
4.1 Reach-Avoid Problem in Joint State-Space
4.2 SD HJ Reachability in the Joint State Space (Oine and Online)
4.3 Mitigating the curse of dimensionality using Hamilton-Jacobi projections
4.4 Simulations
Chapter 5 Extension to C-space and case study with the 2-Do F robot arm
5.1 Problem formulation
5.2 Reach-Avoid Problem and SD HJ Reachability in the CT -space
5.3 Projective under-approximation of the SD reach-avoid tube
5.4 Case study with the 2-Do F robot arm
Chapter 6 Conclusion and Future Prospects
6.1 Summary of results
6.2 Future prospects
Acknowledgements
Publications
Bibliography
本文编号:3524349
本文链接:https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/3524349.html