Project 1: Specification for Automail
Background: Automail
Delivering Solutions Inc. (DS) has recently developed and provided a Robotic Mail Delivery system
called Automail to the market. Automail is an automated mail sorting and delivery system designed to
operate in a large building that has a dedicated mail room. The system offers end-to-end receipt and
delivery of mail items within the building and can be tweaked to fit many different installation
environments.
The current version of the system supports delivery of letters using one specific delivery mode where
every robot operates with the same behaviour. DS would like the system to also handle parcel
delivery, and to support exploring alternative delivery modes which specialize the robot behaviour.
The Automail System
The building the system operates in (see Figure 1) will have a number of floors each with the same
number of rooms. Below the rooms is a mailroom (row 0), to the left and right are robot tracks (column
0 and column roomsperfloor+1) respectively.
Figure 1: Automail Building Layout (6 floors by 5 rooms) with Robots
The mail items are letters which are effectively weightless, or parcels which have a given weight. All
mail items are addressed to deliver within the building to an room identified by combination of Floor
and Room number. The current system handles only delivery of letters.
The Automail system consists of two key components:
A MailRoom subsystem which holds mail items after their arrival at the building’s mail room. The
mail roomdecides the order in which mail items should be delivered.
Delivery Robots which take mail items from the mail room, or from other robots, and deliver them
throughout the building. The currently used robot (see Figure 2) has a backpack-like container for
carrying mail items. Carrying capacity of the robot is weight limited. As such the total weight of
carried items cannot exceed robot.capacity. An installation of Automail can manage a team of
delivery robots of configurable size.
.Figure 2: Artistic representation of one of the DS robots
DS provides a simulation subsystem to show that Automail can operate to deliver mail items within the
building. The subsystem runs a simulation based on a property file, generates an output trace of
execution and outputs the average time to deliver a mail item.
The simulation subsystem uses a clock to simulate operations of the mail room and robot subsystems.
Broadly speaking, for each tick of the clock (i.e. one unit of time), the mail room subsystem will load
items to a robot if there are robots available at the mailroom; and the robots will either move to deliver
an item (if there are items in their backpack), deliver an item, or move to return to the mailroom (if all
items are delivered). Currently, the robots offered by DS will take one unit of time when
moving one step (i.e., moving up or down one floor in a building, or left or right one across the
building).
delivering items at one apartment once there.
transferring items from one robot to another.
being loaded and moved out of the mailroom.
The simulation completes only after all items have been delivered.
Unfortunately, the staff who designed and implemented the Automail simulation did not consider
maintainability and future enhancements.
Current operating mode: Cycling
This mode has been implemented in the current simulation. It has the robots moving clockwise
through the building delivering mail items, as illustrated in this video
(https://canvas.lms.unimelb.edu.au/media_attachments_iframe/20429677?type=video&embedded=true) .
The Cycling mode (MODE=cycling) has every robot operating with the same behaviour. All items a
robot carries for delivery will only ever be for a single floor at a time.
Initially: All robots (robot.number>0) start in the MailRoom.
Ongoing (each timestep):
1. If there are any items and robots in the mailroom, the mailroom will Load a robot with items for one
floor (the floor with the earliest remaining delivered item), and move it to the bottom left (0, 0).
2. If a robot has items to deliver and has not reached its target floor, it will Step up.
3. If a robot has an item to deliver it will Step towards the left-most location to which its items are
addressed.
4. If a robot has reached its delivery location and not yet delivered, it will Deliver.
5. If a robot has no items to deliver, it will Step towards the bottom right (0, building.roomsperfloor+1).
6. If there is a robot at bottom right (0, building.roomsperfloor+1), it will Return to the mailroom.You can assume that the Automail Cycling mode has been well tested and performs reasonably well,
though only for letters.
Proposed operating mode: Flooring
This mode has not been implemented in the current simulation; you and your team need to add this. It
involves having one robot per floor delivering on that floor, with two robots (one at each end) bringing
mail items up to the other robots for delivery, as per this video
(https://canvas.lms.unimelb.edu.au/media_attachments_iframe/20429779?type=video&embedded=true) .
The Levels mode (mode=FLOORING) always involves exactly building.floors+2 robots
one for each floor, each of which exhibits floor behaviour, and
one for each of the leftmost and rightmost columns, each which exhibits column behaviour.
Floor robots only ever move on their floor and column robots only ever move on their column and in/out
of the mailroom.
Floor Behaviour:
Initially: the robot is at Room 1 on their floor.
Ongoing (each timestep):
1. If the robot has mail items, continue delivering them (ignore column robots).
2. If the robot is next to a waiting column robot (i.e. with items for this floor), Transfer them from
the column robot to this robot and start delivering from that end of the floor towards the other
end. (Note: as all robots have the same capacity and the floor robot is carrying nothing, all items
will be transferable.)
3. If the robot is heading for a waiting column robot, continue moving towards that robot.
4. If a column robot is newly waiting, start heading towards it. If two column robots are newly
waiting, move towards the one with the earliest arrival item, or the left one if the arrival time is
the same.
5. If the robot has no items and no column robots are waiting for this floor, then do nothing.
Column behaviour:
Initially: the robot is in the mailroom and is assigned to the left or right column (one robot to each).
Ongoing (each timestep):
1. If the robot is in the mailroom and the mailroom has items, Load the robot (as per Cycling) and
move it to its floor 0, otherwise do nothing.
2. If the robot is loaded for delivery, and not at the destination floor move towards the destination
floor.
3. If the robot is at the destination floor, and the floor robot is adjacent, Transfer from this robot to
the floor robot.
4. If the robot has transferred, head towards floor 0.
5. If the robot arrived back at floor 0, enter the mail room.
All items a robot carries for delivery will only ever be for a single floor at a time.
Robot Operations:
Step: move one floor or apartment towards the destination. Only one robot can be on a square at a
time. If the destination square is blocked, the robot does nothing.Load: the load destination is the floor of the item in the mailroom with the earliest arrive time. Load
the robot with all items destined for that floor which can be transferred to the destination robot,
subject to load. All letters are transferred, and parcels are transferred in order of "earliest arrival
time package which will fit within weight capacity".
Transfer: all items which can be transferred are move from the source robot to the destination
robot, subject to load. All letters are transferred, and parcels are transferred in order of "earliest
arrival time package which will fit within weight capacity".
Return: all items are transferred from the robot back to the mailroom.
Deliver: the item is removed from the robot and the elapsed time since arrival used in the item
statistics.
Your Task
To expand the usage and trial different modes of operation, DS wants to update their Automail to
support (1) parcels, and (2) the FLOORING operating mode.
The Base Package
You have been provided with a zip file containing source code for the current version of the
system, including an example property file.
(https://canvas.lms.unimelb.edu.au/courses/187398/files/20444343?wrap=1)
(https://canvas.lms.unimelb.edu.au/courses/187398/files/20444343/download?download_frd=1) This
provides the basis for you to implement the additions described above.
Please carefully study the provided code and ensure that you are confident you understand how it is
set up and functions before continuing. Note that you do not need to understand all aspects, just those
relevant to the changes you need to make. If you have any questions, please make use of the
discussion board.
Note: The simulation will run and generate mail items at random times and with random weights,
based on a seed. You can configure this in the property file (test.properties by default). Any integer
value will be accepted, e.g. 30006.
Configuration and Project Deliverables
(1) Extended Automail: As discussed above, and for the users of Automail to have confidence that
changes have been made in a controlled manner, you are required to preserve the Automail
simulation’s existing behaviour. Your extended design and implementation must account for the
following:
Preserve the existing behaviour of the system for configurations where the additional capabilities
are turned off in the configuration file (properties), i.e. mail.parcels=0 and MODE=cycling. Note that
“preserve” implies identical output. We will use a file comparison tool to check this.
Add the handling and delivering behaviour for parcels (including robot capacity limitation).
Add the new FLOORS mode of delivery.Configurable building size and number of robots (robot.number for cycling mode or
building.floors+2 for flooring mode).
It's recommended that you understand the high-level design of current system so that you can
effectively identify and update relevant parts. You don't need to refactor the whole system, just those
parts necessary or helpful to making the required changes.
(2) Report: In addition to the extended Automail, DS also wants you to provide a report to document
your design changes and justification of your design. You should also comment on how easy your
changes make it to add further mail items (beyond letters and parcels), or further delivery models
(beyond cycling and flooring) in the future. Your report should include:
a design class diagram which shows all of the changed design elements in your submission (at
least - it can show more than just the changes but doesn't need to show all unchanged elements).
a sequence diagram which illustrates the behaviour of a floor-assigned robot in FLOORING mode,
for appropriate scenario of your choosing.
More detail of the report is available on the LMS submission page.
Note: Your implementation must not violate the principle of the simulation by using information that
would not be available in the system being simulated. For example, it would not be appropriate to use
information from the simulation package (e.g., mail items which have not yet been delivered to the mail
room). We also reserve the right to award or deduct marks for clever or very poor code quality on a
case-by-case basis outside of the prescribed marking scheme.
Testing Your Solution
We will be testing your application programmatically, so we need to be able to build and run your
program without using an integrated development environment. The entry point must remain as
“Main.main()”. You must not change the names of properties in the provided property file or require the
presence of additional properties.
Note: It is your team’s responsibility to ensure that the team has thoroughly tested their software
before submission.
Here are sample outputs for each of the two provided property files:
test.properties (https://canvas.lms.unimelb.edu.au/courses/187398/files/20494343?wrap=1)
(https://canvas.lms.unimelb.edu.au/courses/187398/files/20494343/download?download_frd=1)
testShort.Properties (https://canvas.lms.unimelb.edu.au/courses/187398/files/20494342?wrap=1)
(https://canvas.lms.unimelb.edu.au/courses/187398/files/20494342/download?download_frd=1)
Submission
Detailed submission instructions will be posted on the LMS. You must include your team number in all
your pdf submissions, and as a comment in all changed or new source code files provided as
相关推荐
从零搭建Arduino仿真开发环境:SimulIDE与IDE配置全攻略
本文详细介绍了如何从零搭建Arduino仿真开发环境,涵盖SimulIDE与Arduino IDE的配置全流程。通过仿真模拟环境,开发者可以零成本试错,快速验证电路设计和程序逻辑,特别适合初学者和便携开发需求。文章提供了软件安装、环境配置、项目实战及问题排查的完整指南,帮助读者高效掌握Arduino仿真开发技巧。
Jackson 序列化 / 反序列化注解 完整版笔记
Jackson 序列化 / 反序列化注解 完整版笔记
大华事件检测智能服务器,大华股份:高速公路智能事件检测 解决方案
1方案背景随着我国基础设施建设的飞速发展,公路尤其是高速公路的总里程数年年攀升。交通运输部统计数据显示,截止2017年底全国公路总里程477.35万公里,其中高速公路13.65万公里,位居世界第一。高速公路的运营和管理作为全国交通管理的重要组成部分日益受到严峻挑战,交通拥堵和事故频发已成为高速公路上非常突出的问题。抛洒物、行人、车辆不规则行驶等事件是导致以上结果的重要原因。基于视频分析的高速公路智...
Web 版翻译者资料库
本项目是桌面版《翻译者资料库》的 Web 移植版,基于 Spring Boot 3.3、Thymeleaf、Bootstrap 5.3 与 MySQL 8.0 构建。它面向翻译写作提供资料、词汇、句库、文本工具、对照制作、查找、备份恢复及旧库迁移等功能,完整继承桌面版核心业务规则,支持一键启动与离线访问。
综合能源系统多时间尺度优化调度!诸多创新点让你抓住审稿人的心!
主要介绍综合能源系统的设备组成结构,并针对屋顶光伏系统、冷-热-电联供系统中的燃气轮机、燃气锅炉、光伏溴化锂制冷机、卡琳娜余热发电、中央空调系统,以及外网交互系统的配电网与天然气管网交互关系进行数学建模。1) 源-荷-储协同调度:研究屋顶光伏、燃气轮机、燃气锅炉等源侧设备,中央空调、电-热-冷负荷、可平移电负荷、可削减温控负荷等负荷侧资源,以及储能系统、电动汽车等储能侧设备的协同调度,以提高风光消纳能力,满足社会能源需求。通过电转热、电转冷、电转气等多能耦合方式,可进一步促进可再生能源的应用与发展。
一个固定翼飞机仿真程序带你系统性上手如何实现配平(工具箱+脚本)
老规矩,该博客所用的模型和脚本已上传至百度网盘,名字为“固定翼配平训练例子”,自己自取。
MCP 协议会成为前端 AI 工程的标准接口吗?
MCP(Model Context Protocol)正在成为连接 AI 模型与外部工具、数据源的一层通用协议。本文从 MCP 的核心机制出发,分析它在浏览器与前端工程中的落地边界,讨论它能否成为前端 AI 工程的标准接口,并给出当下更务实的架构选择建议。
openclaw通俗入门系列9——流式传输 + 分块
流式传输 = 不等写完就开始发。分块流式传输是把长回复切成几条消息发(所有渠道),草稿流式传输是像打字一样实时显示(仅 Telegram)。可以配置块大小、发送时机、合并规则、类人延迟。
LangGraph 本地 build 频频翻车,最终在服务器跑通流式对话
先说结论:我本地那个,一次都没成功过。三次,三次,全挂在拉镜像上。而顺得像刚上过油。这就很像我高中那辆自行车:骑着能走,一上锁就坏。那天晚上我盯着终端里那行红字看了很久,想的不是技术方案,想的是——为什么同一个工具,可以一边让我觉得自己很行,一边让我觉得自己不太行。
2026年AI视频生成大模型全景指南:国内外主流工具深度对比与选型建议
AI视频生成技术在2025年至2026年间取得了突破性进展,从"能用"进化到"好用",商业化应用加速落地。中美双轨竞争格局基本成型,国内模型在技术追赶和本土化体验方面展现出强劲竞争力。对于用户而言,选择合适的AI视频生成工具需要综合考虑技术性能、使用门槛、成本预算和具体需求。零基础新手推荐海艺AI、通义万相等低门槛工具;进阶创作者可根据需求选择可灵AI、即梦AI、海艺AI等;专业开发者和企业用户可考虑智谱清影的开源能力或可灵AI的商业化服务。
【WRFDA教程】客观分析(OBSGRID)详解
客观分析(OBSGRID)详解
带着小白学习-微服务-6 实战:网关
地址栏输入:http://localhost:8088/goods/index,这时候就不是到8100端口去调用了,而是8088直接通过路由转发,分别到nacos对应的goods对应的服务中区调用。** 可以看出,这里将前面已经正常运行的server-store子工程(库存服务,服务提供方),server-goods子工程(商品服务,服务消费方)。(1)uri: lb://store 代表要到nacos的注册中心的服务列表中找到服务名为store的微服务;,网关是做路由转发,不是服务调用;
LightGBM之LGBMRegressor()参数详解以及调参
LightGBMLight Gradient Boosting Machine)是一种梯度提升框架,它使用决策树作为基学习器。LightGBM 可以说是在 XGBoost 上做的优化。LightGBM 为高效并行计算而生,它的 Light 体现在以下几个点上:更快的训练速度更低的内存使用支持单机多线程,多机并行计算,以及 GPU 训练能够处理大规模数据。
海外平台频繁弹验证?从 DNS、WebRTC 到浏览器指纹的环境异常排查清单
在运营多地区海外平台账号时,即使代理节点通畅、TCP 握手成功,仍可能遇到频繁人机验证、地区限制、播放失败等异常。核心结论是:这类问题往往不在网络传输层,而在浏览器环境参数与网络出口之间的逻辑自洽性。本文从 DNS 泄漏、WebRTC 穿透、时区与语言匹配、Canvas/WebGL 指纹、TLS 特征五个维度,给出一份可复现的环境异常排查清单,并对比个人手动配置与工程化环境管理方案的适用边界。适用于跨境运营、多账号运维团队。
WINCC7.3破解文件
WINCC7.3硬件够破解文件,有需要的下载!WINCC7.3硬件够破解文件,有需要的下载!
贝叶斯统计 [韦来生编]
贝叶斯(Bayes)统计是近几十年来迅速发展起来的数理统计的一个重要分支。贝叶斯方法与经典统计方法的主要不同之处是进行统计推断时除了利用样本信息外,还要利用参数的先验信息,因此可以提高统计推断或统计决策的效果。它在经济、金融、生物、医学、自然科学和社会科学等许多领域具有广泛的应用。贝叶斯方法的研究已渗透到了统计学的几乎所有领域。作者在给中国科学技术大学概率论与数理统计专业本科生开设的“数理统计”课程中有一章专门讲授“贝叶斯方法和统计决策理论”,近十年来也给中国科学技术大学概率统计专业研究生开设过几次“贝叶斯分析”课。本书是在对过去讲稿的内容作了适当的增减和调整的基础上完成的。 本书共分六章。第1章是绪论,介绍了贝叶斯统计的若干基本概念,同时对必要的数理统计的基础知识有重点地作了回顾。第二章介绍了确定先验分布的若干可供选择的方法。第三章介绍了几类常见统计模型参数的后验分布的主要结果和计算方法。第四章和第五章分别介绍了“贝叶斯统计推断”和“贝叶斯统计决策”的内容。第六章介绍了贝叶斯统计计算的若干方法,包括蒙特卡洛方法和马尔可夫链蒙特卡洛(MCMC)方法以及相关统计软件的简介。本书的每一章附右大量的例题和习题,书末附有常用的几个表格和部分习题解答供读者查用。 大约可在54学时内讲授本书的第1章至第六章的主要内容,教师可根据学时需要选讲本书第1章至第六章的部分内容。书中标“*”号的章节可略去不讲,留给读者作为阅读材料。第六章例题中的R代码和部分例题的数据文件等可从作者个人主页下载。 作者在编写本书的过程中参考了J.O.Berger教授的《统计决策论及贝叶斯分析》和茆诗松教授编写的《贝叶斯统计》等书中的一些内容,以及《贝叶斯分析》一书中由张伟平博士编写的有关贝叶斯统计计算的一些内容。在此表示衷心感谢。 本书编写准备过程中,中国科学技术大学统计与金融系的几个研究生帮助完成了本书部分中文的录入和排版,作者对他们的辛勤工作表示真诚的感谢。高等教育出版社为本书的出版给予了大力支持。在此一并致谢。
智能问诊的流程
本文介绍基于FastAPI构建的医疗多Agent系统MediAgent,通过症状分析、诊断推理与治疗建议等模块协同,实现从问诊到报告生成的完整工作流。系统采用前后端分离架构,结合MySQL、RabbitMQ与Docker实现工程化部署,支持异步化验结果接入。虽当前状态暂存于内存,但已具备可扩展性,未来可通过Redis持久化、消息队列联动及医疗安全机制完善,应用于慢病管理、辅助决策等场景,推动AI在医疗流程中的落地应用。
数学建模模型总结
这是本人在两年内准备参加数学建模竞赛搜集到的模型资料,已经经过了本人的筛选和整理,希望能够对准备参加数学建模竞赛的学弟学妹们提供一些帮助!
SSD测试工程精通系列 17|TRIM与UNMAP洪峰下的GC竞争(命令风暴与后台任务调度)
本文深入探讨NVMe SSD在面临TRIM/UNMAP命令洪峰时,后台垃圾回收(GC)与前端IO的竞争机制。从协议解析、固件调度、硬件架构到物理层信号,全面剖析命令风暴对QoS和WAF的影响。结合Samsung、Micron、Kioxia等厂商实测数据与失效案例,提供从测试注入、仪器抓取到工程取舍的全链路验证方案,助力资深测试工程师攻克存储性能断崖难题。
复分析 阿尔福斯(Ahlfors) 第三版答案 英文
答案分为十章,有部分习题没有答案,全英文复分析阿尔福斯(Ahlfors) 第三版答案
HTML5 History API 详解:前端路由底层原理
pushState改 URL 不刷新popstate接浏览器导航服务端必须兜底 SPA 入口前端路由的本质不是“魔法”,而是:把「URL 字符串」变成「组件渲染函数」的映射表,再让浏览器历史栈和这张表保持同步。
基于 YOLOv8 的森林火灾检测系统(Web 版)
本项目面向林区早期火情识别场景,使用YOLOv8完成火焰(Fire)与烟雾(Smoke)两类目标的检测模型训练,并以FastAPI构建 Web 检测平台,把模型能力封装为可直接登录使用的系统。平台除常规的图片、批量、视频、摄像头检测外,还提供同图多权重对比、数据集浏览与标注核对、模型热切换,以及把训练搬到网页上的能力:可在浏览器配置超参数发起真实训练,训练中在线调整学习率、查看进度与动态曲线,训练完成后下载产物或直接应用到检测。
身份验证漏洞专题(三):持久登录、密码重置与密码修改中的凭据绑定缺陷
密码登录与双因素认证是身份验证的主干流程,但一个完整的认证体系还依赖三类辅助机制:持久登录("记住我")、密码重置、密码修改。这三类机制的共同特征是:它们同样携带或验证凭据,同样直接决定认证状态的归属,但实现时获得的防护强度往往远低于登录页面本身。攻击者绕不开登录接口时,常常可以从这些辅助入口进入。
2026 年前端框架排行榜
前端框架是用于构建用户界面的工具集合,通过预设组件、网格系统和样式规范提升开发效率与可访问性。2026年主流框架包括:React(生态领先,适合全栈)、Next.js(SSR/SEO优化,主流选择)、Vue(学习友好,中小团队首选)、Angular(企业级稳定)、Svelte与Astro(性能突出)。趋势显示,服务端渲染、TypeScript、AI工具融合及轻量化成为关键方向。就业建议:优先学React+Next.js;追求易用选Vue;企业级项目可学Angular;追求极致性能可尝试Svelte或Astr
Vue 前端开发 AI 对话页面:流式消息渲染、文件上传
关键点是不能手动设置 Content-Type 为 multipart/form-data,必须由浏览器自动生成带有 boundary 的请求头,否则后端无法解析。思路:将对象序列化为 JSON 字符串,作为 FormData 的一个普通字段(如 formData.append(‘payload’, JSON.stringify(data))),后端接收该字段后再反序列化。若上传大文件(>10MB),需实现独立的分片上传接口,在提交对话前先获取文件 id 或 url,再随文本发送引用。移除输入框的禁用态;
前端要加密手机号,密钥却不能下发:一次一密的 ECDH + HKDF 方案(十六)
前端加密手机号需避免密钥暴露,采用ECDH协商临时密钥,结合HKDF派生并绑定协议上下文,使用AES-GCM实现认证加密。关键在于两端参数严格一致:曲线、编码格式、盐值与info字段逐字节相同。避免直接哈希共享秘密或日志泄露密钥,服务端返回临时公钥并签名,前端先验签再派生密钥,确保前向安全与防中间人攻击。最终密文结构包含公钥、nonce、密文与认证标签,支持自校验。
浏览器端素材压缩与上传实践:从 Canvas 到 WebCodecs 与 FFmpeg 回退
本文介绍浏览器端素材上传的完整处理链路,重点解决图片与视频超限问题。图片通过Canvas等比缩放与质量压缩,逐级优化至2MiB内;视频则优先使用WebCodecs硬件编码,结合分辨率降级、码率预算与FFmpeg WASM回退,确保在20MiB内完成转码。强调“先确认后处理”原则,避免无谓资源消耗,并通过多层能力检测保障兼容性与性能。核心思想:压缩成功 ≠ 上传成功,需独立校验。
Geejing WebBuilder AI SQL:一句话生成查询的完整链路与「生成不执行」边界
企业里真正卡住数据核对的,很少是「不会写 SQL」。更常见的是三件小事叠在一起:这张表到底叫什么、日期字段是 `DATE` 还是字符串、当前这套库分页该用 `LIMIT` 还是别的写法。一个熟手要先翻一遍表结构,生手干脆把需求丢给 DBA,一来一回半天就过去了。
从一次检测到持续验证,重新认识 Web 应用安全测试
围绕漏洞修复后再次出现的问题,结合 Burp Suite DAST 与 GraphQL 示例,解析周期检测、证据管理和整改复测的实现思路。
副队长ES6+教程(3)--解构赋值
解构赋值是JavaScript中高效提取数组、对象数据的语法,支持按位置(数组)或属性名(对象)匹配,可设置默认值、跳过元素、使用剩余运算符。在Vue开发中广泛用于解构props、响应式数据及函数参数,提升代码简洁性与可读性。掌握其规则与常见陷阱,能显著提高开发效率。
526



被折叠的 条评论
为什么被折叠?



