首页 > 其他分享 >CS823 智能代理

CS823 智能代理

时间:2023-07-09 10:33:16浏览次数:32  
标签:domain CS823 houses house 代理 智能 location model your


CS823 – Reasoning for Intelligent Agents
Computer and Information Sciences
University of Strathclyde
Created June 2023
By Dr Andrew Abel
Redit Assignment – PDDL Building – Worth 100% of Overall Mark
There are 3 parts to this assignment, a non-temporal model, a temporal model, and a set of written
questions.
Submission deadline: 04/08/23
Online submission through MyPlace
Background: Island Construction
You have been hired to model and create automatic plans for a construction company. Island Construction
have a new project, specialising in constructing houses on the mysterious Isla de Planificación. Your task is
to use PDDL to model their work environment and all the agents within it, and to create valid plans. Isla de
Planificación is shown below:
All houses require raw materials, and they can be found in the following depots:
(0,0) Bricks
(0,1) Wood
CS823 – Reasoning for Intelligent Agents
Computer and Information Sciences
University of Strathclyde
Created June 2023
By Dr Andrew Abel
(4,2) Glass
(2,0) Slate
The other locations on the map can potentially have houses built on them, depots can not have houses
constructed on them: (1,2) (2,1) (2,3) (2,4) (3,2) (3,3) (4,3).
Houses require the following resources to be constructed:
1 x Bricks, 1 x Wood, 1 x Glass, 1 x Slate.
There are a number of agents in the company. The company has 2 delivery trucks, which can travel to any
location or depot, and can each carry 1 item at a time, 1 architect, who designs the house, and 1 builder,
who constructs the house. The architect and builder can only travel to locations, not depots, and all agents
must follow the roads to travel between locations. The roads are all two way.
The architect can design the house, but only when there is 1 x wood on the site. The architect can
travel to any location, but not a depot.代写书CS823程序  Designing the house does not use any wood. The builder
starts at location (3,3).
The builder can build the house, but only when there is a designed house on the site. The builder
can travel to any location, but not a depot. When the house is built, all resources are used up. The
builder starts at location (1,2).
The delivery trucks can travel to any depot or location. They can store any type of item, but can only
store 1 item at a time (i.e. they can only carry a single wood). They can load at a depot, or unload at
a location. All delivery trucks start at location (2,1).
We make the assumption that any number of agents may occupy a location or depot simultaneously, and
that the depots produce an unlimited number of resources.
Part 1: Non-Temporal Plan (35 marks)
Use the above information to create a PDDL file for the domain and the problem.
You should correctly identify the agents, functions, and program the actions with suitable conditions. Here,
you do not need to worry about temporal actions.
The goal is to construct houses at (4,3), (1,2), (2,3), creating 3 houses in total:
(=(built_houses loc_43) 1)
(=(built_houses loc_12) 1)
(=(built_houses loc_23) 1)
Note, it may take a minute or two to find a plan, depending on your computer. You should submit your
completed problem and domain files using the names _studentnumber_part1_problem and
_studentnumber_part1_domain respectively.
Part 2: Temporal Planning (35 marks)
In Part 2, you will improve on your initial model by applying temporal information to make temporal plans.
All of the information from part 1 is valid, and some additional temporal information is supplied below. You
CS823 – Reasoning for Intelligent Agents
Computer and Information Sciences
University of Strathclyde
Created June 2023
By Dr Andrew Abel
should use this to create a problem and domain file in pddl. The goal is the same as part 1. You should
produce a plan which is calculated in minutes.
Additional Information:
Distances are the same in both directions:
(0,4) – > (2,4) – 18km
(1,2) - > (0,2) – 7km
(3,2) - > (4,2) – 4km
(4,3) - > (4,2) – 5km
(2,1) - > (2,0) – 5km
(1,2) - > (2,1) – 4km
(1,2) - > (2,3) – 8km
(2,1) - > (3,2) – 3km
(2,3) - > (2,4) – 9km
(2,3) - > (3,3) – 6km
(3,2) - > (3,3) – 7km
(3,3) - > (4,3) – 5km
Due to the condition of the road, all speeds on the island are limited to 20km/h. There is also time
information. It takes time to load the trucks at the depots:
Loading times: brick - 60 mins, wood – 40 mins, glass 30 mins, brick – 60 mins
All unloading takes 20 minutes regardless of material
It takes the architect 50 minutes to design a house, and the builder 70 minutes to construct it.
Note, it may take a minute or two to find a plan, depending on your computer. You should submit your
completed problem and domain files using the names _studentnumber_part2_problem and
_studentnumber_part2_domain respectively.
Part 3: Model Analysis (30 marks)
Using the model you created in part 2, and assuming the goal is to build 1 houses in locations (4,3), (1,2),
(2,3), creating 3 houses in total:
(=(built_houses loc_43) 1)
(=(built_houses loc_12) 1)
(=(built_houses loc_23) 1)
answer the following questions. All questions should be answered in depth, with a discussion of how the
options affect your model, as well as showing what changes you made to the model to investigate the
question.
CS823 – Reasoning for Intelligent Agents
Computer and Information Sciences
University of Strathclyde
Created June 2023
By Dr Andrew Abel
3.1 Vehicle improvements (10 marks)
Management has asked you for your recommendation regarding vehicle improvements. They have a choice
of either improving the existing vehicles (builders, architects, trucks) so that they go 20kmh faster, i.e.
doubling the speed from 20 to 40, or adding an extra truck. With reference to your model, and showing your
working, give your recommendation.
3.2 Staff Training (10 marks)
Management has asked you for your recommendation regarding training. They have two possible training
programmes that may improve performance and would like your advice of which one would be better. They
can either training the loading staff to reduce all loading times by 20 minutes, or they can train their builders
to reduce house building time by half and also use standard designs so that the architect’s design time is
reduced to only 5 minutes per house. With reference to your model, and showing your working, give your
recommendation.
3.3 Infrastructure Improvement (10 marks)
The local government have also proposed improvements. They have proposed to build a new road. The
choices are:
1) Build a single highway from the brick depot at location (0,4) to location (2,3) of length 10km,
2) Build 2 smaller highways one from (2,3) to (3,2), and the other from (1,2) to (3,2), both of length
3km.
They have asked for your advice as to which will improve the speed of house construction. With reference
to your model, and showing your working, give your recommendation.
Submit your answers as a pdf file and also attach any supporting PDDL files you need.
Submission Instructions
You must submit a minimum of 5 files:
1. the domain and problem files for part 1
2. the domain and problem files for part 2
3. a pdf file for part 3 (along with any other problem or domain files you feel are needed)
Use your student id number for the file name, and then label as part 1, 2, or 3 as appropriate.
Additional Notes
Parts 2 and 3 will likely require the use of the Optic Planner on a linux machine that you needed to use
during the semester. Please see the guidance on MyPlace for installing this.
WX:codehelp


标签:domain,CS823,houses,house,代理,智能,location,model,your
From: https://www.cnblogs.com/javanew/p/17538399.html

相关文章

  • 用 AIGC 重构后的智能客服,能否淘到大模型时代的第一桶金?
    ChatGPT的诞生打响了现代AI军备竞赛的第一枪。以GPT-4、ChatGTP、Bard等为代表的大语言模型在全球各界引起了广泛关注。结合ChatGPT的底层技术逻辑,未来中短期内ChatGPT产业化的方向大致有四类:即智能客服、文字模态的AIGC应用、代码开发相关工作以及图像生成。其中,最适......
  • 基于AidLux的自动驾驶智能预警应用方案
    基于AidLux的项目实战之智能预警在AidLux上的部署与应用1.YOLOP模型onnx转换部署YOLOP导出onnx模型执行命令:python3export_onnx.py--height640--width640执行完成后,会在weights文件夹下生成转换成功的onnx模型 2.AidLux模型转换工具AlModelOptimizer通过该工具对onnx......
  • 微信小程序websocke协议wss用nginx反向代理
    map$http_upgrade$connection_upgrade{defaultupgrade;''close;}upstreamwebsocket{serveryourip:port;}server{listen443;server_namexxxx.com;sslon;ssl_certificate/xx/xx/cert.crt;ssl_certificate_key/xx/xx/cert.key;......
  • http 和 https区别,自动携带cookie的session对象,响应response,下载图片视频到本地,编码
    1.1自动携带cookie的session对象#session对象----》已经模拟登录上了一些网站---》单独把cookie取出来 -res.cookies-转成字典res.cookies.get_dict()#请求头和数据importrequestsheader={'Referer':'http://www.aa7a.cn/user.php?&ref=http%3A%2F%2......
  • 智能电视APP鲜时光,如何应用AB测试打造极致的用户观看体验?
     更多技术交流、求职机会,欢迎关注字节跳动数据平台微信公众号,回复【1】进入官方交流群 数字技术的发展让智能电视普及率大幅提升,2023年智能电视的市场渗透率已超90%,与智能电视相匹配的各类应用APP功能也愈加丰富。本文介绍面向智能电视及机顶盒的互联网电视应用——鲜时光A......
  • .NET程序-配置使用Fidder代理,抓取后台HTTP请求
    一般我们用Fidder查看浏览器的Http请求来分析页面的性能问题,但是如果是后台.Net发起Http请求怎么看?下面介绍一种方法,通过配置.NET程序,使用Fidder代理可以查看后台的HTTP请求。首先开启Fiddler2代理: 下面有三种方式设置System.Net使用Fidder作为。方式一:代码方式vardefaul......
  • vue项目打包,配置nginx反向代理
    1.将Vue项目打包在终端中执行npmrunbuild打包完成后可以在项目的dist目录看到index.html(如下图)。将此目录的所有文件放到nginx服务器即可访问。 2.下载nginx并放到服务器上http://nginx.org/en/download.html通过点击上面的链接,下载对应于自己系统的nginx,这里使用Windo......
  • 基于Aidlux的自动驾驶之智能预警部署
    YOLOP能同时处理目标检测、可行驶区域分割、车道线检测三个视觉感知任务,并速度优异、保持较好精度进行工作,代码开源。它是华中科技大学---王兴刚团队,在全景驾驶感知方面提出的模型。这是实操视频:https://www.bilibili.com/video/BV1LX4y1i7mi/?vd_source=4b36e62d12ccafa4305abaa......
  • IM即时通讯+智能审批,助力企业效能提升
    在企业运行中,大都会使用即时通讯去进行沟通交流,即传递文字信息、各类文档文件甚至是实时的语音和视频通话。而随着技术的不断深入,即时通讯软件越发成熟,它已经不再仅仅只是一个简单的聊天工具,而是包涵办公、协作、客户服务等多功能一体化的综合实用平台。而在日常的办公中,企业不仅有......
  • 如何保障业务稳定性?一文详解蚂蚁业务智能可观测平台BOS
    随着业务规模的不断扩大以及AI、云计算、大数据等技术的不断发展,大量的企业希望利用上云来加速其数字化转型,全面提升可靠性、安全性和灵活性,并且降低运营成本。不过对于大多数企业来说,全面上云是一项颇具难度的挑战。这里面原因有很多,无论是复杂遗留系统的迁移难度大,还是数据安全性......