原文传递
Value of Information in Multi Attribute Decision Making for Autonomy.
| 题名： | Value of Information in Multi Attribute Decision Making for Autonomy. |
| 作者： | Kassoumeh, S.; Pandey, V.; Gorsich, D.; Jayakumar, P. |
| 关键词： | Value of information, Multi-attribute decisions, Uncertainty mitigation, Risk aversion, Autonomous vehicles |
| 摘要： | This work presents some results in the value of information calculations for multi-attribute decision making under uncertainty. Almost all engineering activities are undertaken in the face of uncertainty and a decision that maximizes a suitably chosen metric is generally selected. It becomes essential sometimes to collect information regarding these uncertainties so that better informed decisions can be made. Calculation of the worth of this information (VoI) is a difficult task, particularly when multiple attributes are present and there exists dependence between the random attributes in the same alternative or across different alternatives. In this paper, closed-form expressions and numerical models for the calculation of VoI are presented. Particularly, we derive methods for the general scenario where we have to decide over two or more alternatives, each involving two or more continuous random attributes exhibiting some level of dependence with the others. These reduce or completely eliminate the need for conducting simulations or approximations, both of which tend to be either computationally expensive (such as Monte Carlo), limited in accuracy or both. It also allows us to conduct more involved analyses such as sensitivity analysis on design parameters and the engineer鈥檚 preferences in a feasible and even potentially automated way. We also introduce 鈥渁ttribute-wise VoI鈥? which shows that collecting information on one or more of the attribute(s) makes sense only in specific dependence scenarios and trade off relationships between attributes. Calculation methods for value of such information are also provided. We illustrate our models on mobile autonomous system selection decisions. We conclude with a discussion on the avenues for future research into the optimal mix of a system鈥檚 intelligence (autonomy), communication and information gathering. |
| 报告类型： | 科技报告 |
相关文献
- Synthesis of the "State-of-the-Practice for Advancing Planning and Operations Integration Opportunities within Transportation Agencies", Summary.
- Synthesis of the "State-of-the-Practice for Advancing Planning and Operations Integration Opportunities within Transportation Agencies".
- Field Deployment to Quantify the Value of Real-time Information by Integrating Driver Routing Decisions and Route Assignment Strategies. Final Report.
- Exploiting Remotely Piloted Aircraft: Understanding the Impact of Strategy on the Approach to Autonomy.
- Certification Considerations for Adaptive Systems.
- Certification Considerations for Adaptive Systems.
检索历史
应用推荐
